Sie sind auf Seite 1von 114

5

Refraction Surveys

5.1

(1967), Palmer (1980, 1986), and Sjgren (1984); review


papers include Green (1974), Lankston (1990), and
Marsden (1993). Rather than direct the reader to one or
more of these reference books, the main aspects of the
method are presented here. The emphasis is on how it is
used in defining the near-surface layers so that datum
static corrections can be computed. I show that a full
interpretation of refraction data generally requires skill,
ingenuity, and most importantly, exercise of judgment.
Even then, there may be many possible solutions.
A refraction survey acquires information on the time
from the surface down to the refractor and the refractor
velocity. The number of refractors that can be mapped is
generally small because each refractor requires, for optimum definition, the recording of data from a different
offset range. Table 5-1 lists a simplified version of the
information obtained from a refraction survey compared with that from a reflection survey.
In this chapter, Section 5.2 describes basic refraction
theory. This is followed by an analysis of refraction
arrival times (as timedistance curves) for several simple models in Section 5.3. Refraction data can either be
acquired in the field by a separate refraction crewa
weathering or low-velocity layer (LVL) crewor it can
be extracted from the first arrivals recorded as part of a
seismic reflection survey. These two approaches are
compared and their limitations discussed in Section 5.4.
In recent years, the picking of refraction data has
moved on from the classic approach of hand-picking to
methods using computers, either by the field crew or in
a processing center. Various methods are described in
Section 5.5, along with a discussion on the need for quality control (QC) displays to identify regions of poor or

INTRODUCTION

The introduction to Chapter 4 (on upholes) states that


the computation of datum static corrections requires the
definition of the near-surface layers, their thicknesses,
velocities, and variation along the line. The near-surface
detail required is related both to the objectives of the
seismic survey and to the complexity of the near surface.
The thicknesses and velocities of near-surface layers and
the near-surface geology can be estimated with an
uphole survey. However, as pointed out in Chapter 4,
this information is obtained only at discrete points along
the seismic line such that it is necessary to interpolate
between these control points. Interpolation between
uphole survey locations can be based on one or more of
the following: reflection data, refraction data, geologic
data, or simple numerical interpolation; the general
topic of interpolation was discussed in Section 3.4. In
this chapter, I describe the refraction technique both as a
means to interpolate between uphole locations and how
it can be used without uphole control. The various
assumptions and limitations of both approaches are discussed.
Over the years, much has been written about refraction surveys; the seismic refraction technique was used
in the 1920s prior to the introduction of the reflection
method in the early 1930s. Geophysical textbooks such
as Heiland (1940), Nettleton (1940), Jakosky (1950),
Dobrin (1976), Dix (1981), Sheriff and Geldart (1982,
1995), Telford et al. (1984, 1990), Dobrin and Savit (1988),
Sheriff (1989), and Kearey and Brooks (1991) include
descriptions of the seismic refraction method. Books
specifically on seismic refraction include Musgrave

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

141

142

Static Corrections for Seismic Reflection Surveys


Table 5-1. Information Obtained from a Refraction Survey Compared with That from a Reflection Survey.

Item

Refraction Survey

Two-way time information:


Event generated by

Inclined
Velocity contrast

Velocity information
Number of mappable horizons
for a specific offset range
Thin layers
Velocity inversion

Of refractor or just within it

Reflection Survey
Vertical, or more correctly, normal to reflector
Acoustic impedance contrast (product of velocity
and density)
Average down to the reflector

Few at most
Unlikely to be recorded
Not recorded

spurious picks. Prior to the interpretation of refraction


data (the inversion step), various arrival-time adjustments may need to be made to the refraction times.
These are described in Section 5.5.7, where I show that
these may include datum static corrections in some circumstances.
Many different procedures have been proposed for
the interpretation of refraction data, although the differences between some of the methods are fairly small.
Rather than detail just a few, I have mentioned the highlights of most of the methods applicable to near-surface
surveys (Section 5.6). Because many different methods
are currently in use or have been used on data that are
still current, this review can then serve as a complete reference unless more detail is required, in which case the
reader can refer to the original references. The standard
approach of refraction interpretation involves picking
the original, or partially processed, field records. Some
surveys use stacks of the data, such as common source
or receiver stacks, to both enhance the signal-to-noise
ratio and reduce the amount of data to be picked; this
approach is discussed in Section 5.5.
The conversion of refraction arrival times or data into
a near-surface depth profile requires details of the velocity down to the refractor; various approaches and their
limitations are described in Section 5.7. I show that in
many cases interpretive judgment is required, as there is
generally insufficient data available to generate a
unique result. The consequences of this on the depth
profile can be significant, especially when no independent velocity information is available, such as from an
uphole survey. However, Section 5.7.3 shows that, if the
erroneous depth profile is used to compute datum static corrections with the same near-surface velocity profile
used to compute the depth profile, the errors in the
datum static corrections are reduced.
Even with good near-surface velocity control at discrete locations along a line, the interpreter often has a
choice between a smooth near-surface velocity profile
and an irregular refractor depth profile, or a smooth
depth profile and an irregular velocity profile. Two such

Many
Possible to observe
Generates reflected energy if acoustic impedance
contrast exists

models can be generated that will give the same refraction times but different datum static corrections because
of the change in refractor depth. This refractor depth
and near-surface velocity ambiguity is discussed in
Section 5.7.3.1.
Shear-wave refraction surveys are broadly similar to
those used for compressional waves. Minor differences
for shear- and converted-wave surveys compared with
compressional-wave surveys are described in Section 5.8.

5.2
5.2.1

REFRACTION THEORY
Huygens Principle and Refraction
Across an Interface

Refraction can be defined in terms of the change in


direction of a seismic ray or wavefront when crossing an
interface between layers of differing velocities. The simplified description below is based on ray theory and
illustrates what happens when an incident wavefront
approaches an interface between two media. No
account is taken of amplitude and or the three-dimensionality of the wavefield. A more complete description
can be found, for example, in Dix (1939a, b) and Grant
and West (1965).
In a homogeneous medium, the concept called
Huygens principle states that Every point on an advancing wavefront can be regarded as the source of a secondary wave and that a later wavefront is the envelope
tangent to all the secondary waves (Sheriff, 1991).
Thus, when an incident wave strikes an interface separating two media, each point of incidence serves as the
source of a reflected wave in the same medium and as
the source of a refracted wave in the second medium.
The velocity of the medium determines the velocity at
which the waves travel and hence the distance between
successive wavefronts.
Figure 5-1 shows a series of incident wavefronts striking a horizontal interface between two media. The construction of several wavefronts within the upper layer is

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Velocity V1
S

In c i

de

av
nt w

e fro

n ts

t
t + t
t + 2t

1
B
A
2

t + 3t

sin c =
C

ed
ct
ra
f
e
R

the interface, that is, when 2 = 90, the refracted ray in


effect travels along the interface between the two media.
The corresponding incident angle (1) is called the critical angle (c) and is defined as

1
2

Velocity V2

nt
fro
e
av
w

Fig. 5-1. Refraction of plane waves across a horizontal


interface (V2 = 2V1). Variables: 1 is incident angle, 2 is
refracted angle, and t is wavefront separation.

shown, starting from time t with an increment of t. The


distance between these wavefronts is V1t, where V1 is
the velocity in the upper medium. For simplicity, the
incident wavefronts are shown as planes; this is a reasonable approximation over a short segment of the
wavefront if the original source location is a considerable distance from the wavefronts.
The construction of two wavefronts at times of t + 2t
and t + 3t in the lower medium are also shown in
Figure 5-1; this has a velocity (V2) twice that of the upper
medium. The angle between the incident wavefront and
the interface is 1 and for the refracted wavefront the
angle is 2. The positions of the reflected wavefronts and
wavefronts associated with converted waves are not
shown. Using simple trigonometry, we have
sin 1 =

BC V1t
=
AC AC

and
sin 2 =

AD V2 t
=
,
AC
AC

where t is the time separation between successive


wavefronts. These equations can be combined to give
sin 1 V1
.
=
sin 2 V2

(5.1)

Equation (5.1) is the mathematical expression of Snells


law, which is also referred to as Descartes law (Sheriff,
1991). When the refracted wavefront is perpendicular to

143

V1 .
V2

(5.2)

For equation (5.2) to be valid, the velocity in the lower


medium (V2) must be greater than that in the upper
medium (V1). The same expressions hold for the incident and refracted rays, perpendicular to the wavefronts, when the angles are defined with respect to normal to the interface. In Figure 5-1, SA is the incident ray
and ADE is the refracted ray.

5.2.2

Transmission of Refracted Waves

Section 5.2.1 showed that an incident ray or wavefront must strike an interface separating two media at
the critical angle in order for a refracted ray (considered
to be traveling along the interface between the media) to
be generated. This ray and its associated wavefronts
also act as secondary sources for the formation of new
wavefronts in both media. Using a similar approach to
that in Figure 5-1, it can be shown that the emergent
angle from the interface for the wavefront (or ray) in the
upper medium is the critical angle defined in equation
(5.2). The waves that travel to and along the interface
between the two media (the refractor) and return
through the upper medium are referred to as refraction
waves, head waves, Mintrop waves, or bow waves.
Figure 5-2 illustrates wavefronts at equal time increments from a source position at S to receiver positions
R1 and R2 for the simple case of two layers separated by
a horizontal interface, where the velocity of the lower
medium (V2) is three times that of the upper medium
(V1). Figure 5-2a shows wavefronts numbered 1 to 12
associated with the direct wave, which travels in the
upper medium. Figure 5-2b shows the emergent wavefronts (head waves) in the upper medium and the associated wavefronts in the lower medium. Point A on the
interface between the two media is where the incident
wavefront strikes the interface at the critical angle. In
this example, the thickness of the layer was chosen so
that point A corresponds to the intersection of the third
wavefront with the interface. The figure shows that
receiver location R1 corresponds to the point where
wavefront number 8 reaches the surface.
Figure 5-2c illustrates the positions of the wavefronts
in the lower medium and the direct and refracted wavefronts in the upper medium. The dashed line joining A
and R1 represents the positions where these latter two
wavefronts intersect and is sometimes called the coinci-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

144

Static Corrections for Seismic Reflection Surveys


(a)

S
1

10

11

12

Velocity V1
Velocity V2
(b)
S

R2

R1

c
B
6

(c)

R2

R1
2

B
6

Fig. 5-2. Two layers separated by a horizontal interface; numbered wavefront positions from source at S for
(a) direct wave; (b) refracted wave for V2 = 3V1; (c) combined wavefronts. Variables: c is critical angle and R1 is at
crossover distance.

dent-time curve. The direct wave arrives first, to an offset


defined by distance SR1, called the crossover distance. At
offsets greater than this, the refracted waves arrive first
because the time saved by traveling through the highervelocity medium more than compensates for the longer
travel path. The refraction raypath for receiver R2 is
SABR2, where the emergent ray BR2 leaves the interface
at the critical angle (c). The data in Figure 5-2c show
that the apparent velocity of the emergent waves is
given by
Vapp =

V1
.
sin c

Using Snells law, as defined by equation (5.2), we can


transform the above equation to
Vapp = V2 .
Thus, the velocity of the refracted waves measured at
the surface is the velocity of the lower medium.
The shape and attitude of the refracted wavefronts in

Figure 5-2 is the same as some of the wavefronts in


Figures 4-28 and 4-29 which illustrated the wavefield
away from a borehole. The refracted wavefronts were
linear in these two figures and were seen at the far offsets for both layer 1 and layer 2.
The derivation of timedistance curves for simple
two- and three-layer models are shown in Section 5.3;
these include the effects of a thin layer, a velocity inversion, and diffractions associated with a fault. Most
refraction interpretation techniques are based on raypath analysis of the data, as shown in Figure 5-3, which
illustrates a simplified raypath diagram for a two-layer
near-surface model. Graphical techniques, which
require the user to construct wavefronts (e.g., Rockwell,
1967), are described in Section 5.6.8. These approaches
are instructional because, through their use, the interpreter acquires a good understanding of the refraction
technique and the limitations of their data set. In recent
years, the graphical approach has been adapted as a
computer technique in which the wavefronts are downward continued (see Section 5.6.8.5).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

R1

Second arrivals

R12
Slope = __
1
V2

Time

R6

V2

0
xc

xcr

Fig. 5-3. Refracted raypaths for a simple near-surface


model. Variables: V1 is near-surface or weathered layer
velocity and V2 is refractor velocity.

Offset
x
0

R
c

Depth

TIMEDISTANCE CURVES FOR


LAYERED MEDIA

Slope = __
1
V1

t0

V1

5.3

145

Velocity V1
A

Velocity V2

In Sections 5.3.1 to 5.3.4, timedistance curves are


computed for various simple models in which each
layer is assumed to have a constant velocity. Section
5.3.5 provides examples in which a change of velocity
occurs with depth within a layer, which can result from
compaction. Various conditions that preclude the observation of first arrivals from specific layers are described
in Section 5.3.6; these factors often lead to problems in
interpreting refraction data because different models
can produce identical timedistance curves. The impact
of a faulted refractor is described in Section 5.3.7. The
interpretation of refraction observations and timedistance curves is described in Section 5.6, including some
practical problems that occur because the earth is never
as simple as defined by the models described here.

Fig. 5-4. Source-to-receiver raypath and timedistance


curve for two layers separated by a horizontal interface.
Variables: t 0 is intercept time, x is source-to-receiver offset, xcr is critical distance, xc is crossover distance, V1 is
layer 1 velocity, V2 is layer 2 velocity, z is depth to layer
2, and c is critical angle.

5.3.1

Snells Law, as defined by equation (5.2), can be used


to transform equation (5.4) to

Two Layers Separated by a


Horizontal Interface

Figure 5-4 shows the raypath from a source location


at S to a receiver location at R for two layers separated
by a horizontal interface at a depth z with velocities V1
and V2. The total traveltime (tx) for this raypath, with a
source-to-receiver distance of x, can be computed by
summing the traveltime spent on each of the three sections that make up its travel path:
tx =

SA AB BR
+
+
.
V1 V2
V1

(5.3)

Using the symbols defined in Figure 5-4, we can rewrite


this as

tx =

x 2 z tan c
z
z
+
+
V1 cos c
V2
V1 cos c

or
tx =

tan c
x
1
+ 2 z

.
V2
V2
V1 cos c

tx =

(5.4)

sin c sin c
x
1
+ 2 z

V2
V1 cos c cos c V1

or
tx =

x 2 z cos c
+
.
V2
V1

(5.5)

Equation (5.5) represents a straight line with a slope of


1/V2 and an intercept at x = 0 of t0 (referred to as the
intercept time) given by
t0 =

2 z cos c
.
V1

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

(5.6)

146

Static Corrections for Seismic Reflection Surveys

Table 5-2. Critical and Crossover Distances Expressed


as Function of Ratio of Refractor Velocity (V2) to Direct
Arrival Velocity (V1).
Velocity
Ratio
V2/V1

Critical
Distance
xcr

Crossover
Distance
xc

1.2
1.4
1.6
1.8
2.0
2.5
3.0
4.0
5.0
6.0

3.0z
2.0z
1.6z
1.3z
1.2z
0.9z
0.7z
0.5z
0.4z
0.3z

6.6z
4.9z
4.2z
3.7z
3.5z
3.1z
2.8z
2.6z
2.5z
2.4z

tx c =
for the direct arrival, or by

t x c = t0 +

t0 = x c

(5.7)

If the cos c term is replaced with an expression involving the velocities of the two layers, based on equation
(5.2), then equation (5.7) can also be expressed as
z=

t0
V1V2
2
2 V V2
2
1

1/ 2

(5.8)

A refracted wave can be obtained only at offsets


greater than the critical distance. This corresponds to the
source-to-receiver offset when the distance traveled
along the refractor is zero, that is, when points A and B
in Figure 5-4 are collocated and the reflected and refracted raypaths are identical. The critical distance, shown as
xcr in Figure 5-4, can be expressed as
xcr = 2 z tan c .

V2 V1 .
V1V2

Using the relationship between t0 and z defined by


equation (5.7), we can rearrange this to give

The timedistance curve for the model is also shown in


Figure 5-4. The direct arrival, in which the energy travels in the upper layer, is defined by a straight line with a
slope of 1/V1 and an intercept at x = 0 of 0.
Thus, the near-surface layer velocity (V1) and the
refractor velocity (V2) can be estimated from the
timedistance curve. A simple rearrangement of equation (5.6) is used to compute the depth to the refractor (z):
t0 V1
.
2 cos c

xc
V2

for the refracted arrival. These two expressions can be


combined to give

z=

z=

xc
V1

(5.9)

At the crossover distance, xc in Figure 5-4, the direct


arrival and the refracted arrival occur at the same time.
The crossover distance, sometimes erroneously called the
critical distance (Sheriff, 1991), can also be used to compute the refractor depth. At the crossover distance xc, the
traveltime tx is given by

xc V2 V1
.
2 V2 cos c

(5.10)

The approach used in converting equation (5.7) to (5.8)


can now be used to express equation (5.10) as
z=

xc
2

V2 V1

V2 + V1

1/ 2

(5.11)

Critical and crossover distances for a range of velocity ratios are listed in Table 5-2. For example, when the
refractor velocity is twice the near-surface velocity, the
minimum offset required to observe the refractor as a
first arrival is 3.5 times the refractor depth. Between the
critical distance and the crossover distance, the refraction arrival can be observed as a second arrival. This is an
arrival later in time than the first arrival, shown in
Figure 5-4 by a dashed line. The practicalities of using
second arrivals from near-surface surveys are discussed
in Section 5.5.3.

5.3.2

Three Layers Separated by


Horizontal Interfaces

The raypath for a far-offset receiver is shown in


Figure 5-5 for a model consisting of three layers with
velocities V1, V2, and V3 separated by horizontal interfaces at depths of z1 and z2. The timedistance curve for
the first two layers (shown in Figure 5-5) is the same as
that defined in Section 5.3.1 for the two-layer case. The
total traveltime (tx) from source S to receiver R for the
raypath is computed by summing the traveltime spent
on each of the five sections that make up its travel path:
tx =

2 z1
2 h2
x 2 z1 tan 1 2 h2 tan 2
,
+
+
V1 cos 1 V2 cos 2
V3

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

(5.12)

Chapter 5Refraction Surveys

Time

Slope = __
1
V3
Slope = __
1
V2

t02
t01

t0
z cos 1 V2
.
h2 = 2 1
2
V1 cos 2

xc1

xc2
Offset
x

S
1

R
Velocity V1

z1
Depth

The timedistance curve is shown in Figure 5-5 and


includes the direct arrivals and refraction arrivals from
the two layers.
The second layer thickness (h2) can be expressed in
terms of the other parameters by rearranging equation
(5.14):

Slope =__
1
V1

Velocity V2

h2

147

(5.15)

Thus, computation of the depth to the deeper refractor


involves computing the shallower refractor depth and
the thickness of the second layer, followed by a simple
addition to obtain the total depth.
As in the two-layer case, the crossover distances (xc1
and xc2) can also be used to compute the thicknesses and
depths of the layers. For refraction arrivals from both
interfaces to be observed, the velocity must increase
with depth, that is, V1 < V2 < V3.

z2

5.3.3

Velocity V3

Fig. 5-5. Source-to-receiver raypath and timedistance


curve for three layers separated by horizontal interfaces.
Variables: t01 and t02 are intercept times for layers 2 and
3, xc1 and xc2 are crossover distances from layers 1 to 2
and 2 to 3, z1 and z2 are depths to layers 2 and 3, h2 is
thickness of layer 2, V3 is layer 3 velocity, 2 is critical
angle from layer 2 to layer 3, and 1 is incidence angle
within top layer for refracted ray from layer 2 to 3; others
as in Figure 5-4.

Two Layers Separated by a


Dipping Interface

The raypath from a source at S1 to a far-offset receiver


at S2 is shown in Figure 5-6 for two layers with velocities
V1 and V2 separated by an interface dipping at an angle
of . The thicknesses measured perpendicular to the
interface at these two locations are hd and hu, which refer
to downdip and updip recording, respectively.
Using the approach to derive equation (5.4) for the
horizontal layer case (Section 5.3.1), we can compute the
total traveltime for shooting downdip (tx ) as
d

where x is the source-to-receiver offset and h2 = z2 z1 is


the thickness of layer 2.
The critical angle (2) for the refracted ray from layer
2 to layer 3 is given by
sin 2 = V2/V3.
Using Snells Law, equation (5.1), we can give the incident angle within the top layer (1) as
sin 1 = V1/V3.
These relationships and the approach used to derive
equation (5.5) allow equation (5.12) to be expressed as
x 2 z1 cos 1 2 h2 cos 2
tx =
+
+
.
V3
V1
V2

2 z1 cos 1 2 h2 cos 2
.
+
V1
V2

By simple geometry, the recording updip thickness (hu)


can be expressed in terms of hd as
hu = hd + x sin .
Using this relationship for hu, we can rewrite equation
(5.16) as
tx d =

(5.13)

2 h cos c
x
+ d
,
V2 d
V1

(5.17)

where

This represents a straight line with a slope of 1/V3 and


an intercept at x = 0 of t02 given by
t0 2 =

x cos hd tan c hu tan c


hd
+
V1 cos c
V2
hu
+
.
(5.16)
V1 cos c

tx d =

(5.14)

V2 d =

V1
.
sin( c + )

(5.18)

Equation (5.17) represents a straight line with a slope of


1/V2d and an intercept at x = 0 of t0 given by

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

148

Static Corrections for Seismic Reflection Surveys

1
Slope = __
V2d

1
Slope = __
V2u

tr

t0u
1
Slope = __
V1

t0d

where
0 < c .

Time

tr

1
Slope = __
V1
Offset
x

S1

An implication of equation (5.21) is that when > c, the


apparent updip velocity V2u is negative. This means that
arrivals at a long offset occur earlier in time than those
for a slightly shorter offset.
To estimate the refractor depth, we must compute
(c + ) and (c ) from V1, V2d, and V2u using equations (5.18) and (5.21). The critical angle c and dip angle
are then computed from these values. A rearrangement of equation (5.19) allows hd to be computed:

S2

hd =

c
hd

Velocity V 1

Depth

hu

Fig. 5-6. Reversed raypath and timedistance curves for


two layers separated by a dipping interface. Variables: t0d
and t0u are downdip and updip intercept times, V2d and
V2u are apparent downdip and updip refractor velocities,
hd and hu are thicknesses measured perpendicular to the
interface for downdip and updip recording, tr is reciprocal
time, is dip of layer 2; others as in Figure 5-4.

t0 d =

2 hd cos c
.
V1

(5.19)

This is shown on the timedistance curve in Figure 5-6.


Similarly, the total traveltime for shooting updip (tx )
u
can be expressed as
tx u =

(5.20)

V1
.
sin( c )

(5.21)

where

The updip intercept time (t0u) is similar to the downdip


time and can be expressed as
t0 u =

2 hu cos c
.
V1

It follows from equations (5.18) and (5.21) that


V2 < V2 < V2u ,
d

zd =

(5.23)

(5.22)

hd
.
cos

(5.24)

Expressions similar to equations (5.23) and (5.24) apply


to the updip case.
By the principle of reciprocity, the total traveltime from
the updip source to the downdip source and from the
downdip source to the updip source are identical. This
is called the reciprocal time, or occasionally the end-to-end
time, and refers to the raypath from S1 to S2, labeled tr, in
Figure 5-6. For the refraction arrival times to be the same
at these two locations, the relevant sources and receivers
should be collocated with similar ground coupling. In
practice, the source is often located in a borehole; the
uphole time is then added to the arrival times to simulate times from a surface source. A recording geometry
in which source points at each end of the spread are also
occupied by receivers is called a reversed refraction profile.

5.3.4

2 h cos c
x
+ u
,
V2 u
V1

V2 u =

V1
.
2 cos c

The depth (zd) (vertically beneath S1) is then

Velocity V 2

t0 d

Multilayer Case

The timedistance curves computed for a three-layer


case in Section 5.3.2 can readily be extended to a large
number of layers. Equation (5-14) for the three-layer
horizontal case can be rewritten as
t0 2 =

2 h1 cos 13 2 h2 cos 23
+
,
V1
V2

where h1 and h2 are the thicknesses of layers 1 and 2,


with velocities V1 and V2. The angles 13 and 23 are
defined as
sin 13 = V1/V3
and
sin 23 = V2/V3.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

n 1

Vii cos in ,
2h

(5.25)

k = 2.5
k = 5.0
400

V = 600 + kz
k = 10.0

i =1

where Vi is the velocity in layer i and hi is the thickness


of layer i.
Computing the refractor depth is similar to that for
the three-layer case; thicknesses are derived layer by
layer, starting with the top layer. For refraction arrivals
to be obtained from all horizons, the velocity must
increase with depth. At the limit, with very thin layers,
the multilayer case shifts to a continuous change of
velocity with depth (see Section 5.3.5). If one or more of
the layers is not thick enough, refraction arrivals will not
be observed as first arrivals, and the layer is then
referred to as a hidden layer (see Section 5.3.6).
Several authors (e.g., Ewing et al., 1939; Adachi, 1954;
Mota, 1954; Palmer, 1980, 1986) have derived traveltime
formulas for multiple plane-dipping layers. Different
definitions are used by various authors for the angles
and thicknesses; in some cases, the angles are absolute
and in others they are measured with respect to the previous layer.

5.3.5

k=0

Velocity Increase with Depth

In areas where lithology changes slowly as a function


of depth, the velocity may change continuously with
depth as well. Several expressions are used to describe
the variation in velocity with depth:
V(z) = Az1/k,

(5.26)

V(z) = V0ekz,

0
200
Offset (m)

100
V = 600 + 5z

200

Fig. 5-7. Raypaths and timedistance curves for a direct


arrival in a layer characterized by a linear increase in
velocity with depth where V(z) = V0 + k z.

Each definition of the velocity variation with depth


gives rise to different raypaths and timedistance
curves. The Blondeau method, for example, is based on
equation (5.26), in which the instantaneous velocity is
proportional to a power of the depth (Musgrave and
Bratton, 1967) (see Section 5.6.7). Figure 5-7 shows various raypaths and the timedistance curve for the direct
arrival in a layer with a linear velocity increase with
depth, as described by equation (5.27). For this velocitydepth relationship, the depth of greatest penetration
of the ray (zmax) for an offset x is given by the following
(e.g., Dobrin, 1976; Greenhalgh et al., 1980; Greenhalgh
and King, 1981):

V(z) = V0(1 + kz)1/2,


V(z) = V0 + kz,

zmax

(5.27)

and
V(z) = V0 + kz1/2,
where V(z) is the velocity at depth z, V0 is the velocity at
the surface, and A and k are constants.
A gradual increase in velocity with depth implies that
the raypaths are curved. A timedistance curve can
either be analyzed by a succession of linear segments or
by a curved raypath technique (e.g., Ewing and Leet,
1932; Palmer, 1983).

400

Depth (m)

t0 n 1 =

800

Time (ms)

For the general case, in refers to the angle subtended in


layer i for a refraction in layer n. The above expression
for t02 can be extended so that for n horizontal layers, the
intercept time t0n1 is expressed as

149

V
= 0
k

2 2 1/ 2
1+ k x
,

4V02

(5.28)

and the total traveltime along the ray (tx) is given by


tx =

2
kx
sinh 1
,
2V0
k

(5.29)

or

tx =

1/ 2

2
kx
k 2x2

log e
+ 1 +

2V0
.
k
4V02

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

(5.30)

150

Static Corrections for Seismic Reflection Surveys


1
Slope = __
V2

Time

1
Slope = __
V3

t02
t01

1
Slope = __
V1

0
0

Depth

xc2

xc1
Offset

Velocity V1
z1
z2

Velocity V2
Velocity V3

Fig. 5-8. Source-to-receiver raypath and timedistance


curve for three layers, including a hidden layer, separated by horizontal interfaces. Variables as in Figures 5-4
and 5-5.

Figure 5-7 shows that for a value of k = 5, arrival-time


differences between a velocity increase with depth and
a constant velocity (k = 0) are small up to an offset of
about 100 m; this corresponds to a maximum depth penetration of about 10 m. At an offset of 200 m, however,
the time difference is appreciable (about 30 ms); the
maximum depth of penetration is about 35 m, which
equates to a maximum velocity of 775 m/s. Figure 5-7
shows that time differences can be appreciable due to a
combination of factors such as long offsets and large
acceleration values (k). This is only significant in a few
areas because the direct arrival at longer offsets is usually overtaken by a refraction arrival from a deeper layer.
Formulas for other descriptions of velocity as a function of depth and their mathematical derivations are
beyond the scope of this book. Additional information
can be found in most of the references cited in Section
5.1, as well as in Banta (1941), Goguel (1951), Kaufman
(1953), Duska (1963), Evjen (1967), Hollister (1967), Laski
(1973), Greenhalgh et al. (1980), and Greenhalgh and
King (1981).

5.3.6

Hidden Layers, Velocity Inversions,


and Blind Zones

A hidden layer is one that cannot be detected by refraction methods, which may be due to insufficient thickness of the layer or its velocity contrast or to a velocity
inversion. Except for velocity inversions, secondary or
later refraction arrivals can sometimes be used to

resolve a hidden layer. This is because it is only a hidden


layer when the refraction interpretation is limited to
first-arrival information; however, in near-surface surveys, interpretation is generally limited to the first
arrivals. For some authors, a blind zone is synonymous
with a hidden layer (e.g., Sheriff, 1991), while others
have defined it as the maximum thickness of a hidden
layer (e.g., Palmer, 1980). A hidden layer is sometimes
referred to as a masked layer or shadow zone.
If a layer is too thin, refracted energy is not observed
as a first arrival, but it is recorded as a second (or later)
arrival. Under these conditions, a three-layer situation
can easily be mistaken for a two-layer case, leading to
depth values that are too small. This is illustrated in
Figure 5-8 by a modification of the three-layer case (see
Figure 5-5), in which the thickness of the second layer
(velocity V2) is significantly reduced. The resulting
timedistance curve shows that the refracted arrival
from the second layer occurs only as a second arrival.
The crossover distance from layer 2 to layer 3 (xc2) is at
a shorter offset than from layer 1 to layer 2 (xc1). If the
layer 3 depth is computed using only first-arrival information, a near-surface velocity of V1 is used rather than
a higher velocity composed of both V1 and V2, thus
leading to an underestimate of the refractor depth.
The conversion of refraction arrival information into
a depth profile requires the velocity from the surface
down to the refractor. This can be obtained from an
uphole survey or estimated from the refraction survey
itself. In the latter case, if a hidden layer is present but
undetected, the velocity will be in error. The implications of this to the near-surface profile and datum static
corrections are discussed in Section 5.7.
The presence of a velocity inversion causes the wavefronts to be refracted away from the interface so that it is
impossible to have an incident wavefront at the critical
angle. An example of a raypath and the timedistance
curve associated with a velocity inversion are shown in
Figure 5-9. Here, the depth error to layer 3 is opposite in
direction to the thin-layer case (referred to earlier)
because the velocity down to the refractor is too fast
(V1 > V2). This leads to an overestimate of the depth.
Permafrost is a good example of an area where a significant velocity inversion occurs. Here the near-surface
velocity is normally much higher than in the underlying
unfrozen sediments at greater depths. The velocity to a
deep refractor must take this into account to produce a
realistic depth profile.
Thus, for any given timedistance curve, different
interpretations are possible depending on the velocity
distribution present. For example, observed values of a
near-surface velocity of 1250 m/s, a refractor velocity of
2500 m/s, a crossover distance of 260 m, and an intercept time of 104 ms could represent either of the situa-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

151

Table 5-3. Interpretation of Refraction Example.a

Time

1
Slope = __
V3

1
Slope = __
V1

0
xc2
Offset
0

Model 2
(Hidden Layer)

Model 3
(Velocity
Inversion)

1250

1250
1500

1250
900

75

45
39

45
20

75

84

65

Velocity (m/s)
Layer 1
Layer 2
Thickness (m):
Layer 1
Layer 2
Depth to top
of layer 3 (m)

t02

Model 1
(Two Layers)

a This illustrates that three possible near-surface models fit the parameters noted in

the text.

Depth

Velocity V1
z1
Velocity V2 (<V1)
z2
Velocity V3

Fig. 5-9. Source-to-receiver raypath and timedistance


curve for three layers, including a velocity inversion, separated by horizontal interfaces. Variables: xc2 is
crossover distance from layer 1 to 3; others as in Figures
5-4 and 5-5.

tions shown in Figures 5-8 and 5-9. These values correspond to any one of the three models given in Table 5-3
because all three have identical timedistance curves for
first-arrival information. The depth to the refractor
(layer 3) thus varies considerably and depends on the
velocity distribution from the surface to the refractor,
which unfortunately cannot be obtained using refraction first-arrival information. This specific example was
chosen to illustrate the point, but it is not an extreme
case; other models can be constructed to show depth
errors that are significantly greater.
The differences in shapes of timedistance curves
from two directions of recording can sometimes be used
to test for the presence of hidden layers. This is
described in Sections 5.6.5 and 5.6.6, which cover the
generalized reciprocal method (GRM) and the delay
time interpretation technique.
When a hidden layer is caused by insufficient thickness or velocity contrast, its maximum thickness (or
blind zone) occurs when the two crossover distances xc1
and xc2 (see Figure 5-8) are identical. Thus, a first arrival
from the hidden layer occurs at a specific offset (xc1) on
the timedistance curve. In practice, the layer can be
slightly thicker than this because, if only one point on
the timedistance curve corresponds to the hidden
layer, the information is insufficient to define its presence or velocity.

For a three-layer case, the intercept times and velocities required for a correct solution were shown in
Section 5.3.2 to be t01, t02, V1, V2, and V3, or the appropriate velocities and crossover distances. If the second
layer is hidden, the information available is limited to
t02, xc1, V1, and V3. These can be used to derive a depth
to the refractor (z) using the two-layer approach with
one of the equations in Section 5.3.1. True thicknesses h1
and h2 can be determined for an assumed velocity
model and used to give an indication of the maximum
error. For example, if equation (5.11) is used, an incorrect
depth estimate (z) would be
z=

xc 1 V3 V1
2 V3 + V1

1/ 2

When the two crossover distances are identical (xc1 =


xc2), the thickness of the first layer (h1) can be computed
by substituting V2 for V3 in the above equation, assuming V2 is known by another means. This leads to the following ratio of h1 to z:
h1 V2 V1 V3 + V1
=

z V2 + V1 V3 V1

1/ 2

(5.31)

Thus, once the incorrect depth z is computed, the relationship of layer 1 thickness h1 to unknown velocity V2
can be established. An equivalent expression for the second layer thickness can also be computed:
h2 V2
=
z
V1

h V 2 V12

1 1 32

z V3 V22

1/ 2

(5.32)

Nomograms for determining the thickness of blind


zones can be found in Maillet and Bazerque (1931), Leet
(1938), Hawkins and Maggs (1961), Merrick et al. (1978),
Palmer (1980), Schmller (1982), and Won and Bevis

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

152

Static Corrections for Seismic Reflection Surveys

Time

Diffractions

large offsets, however, it can normally be assumed that


the emergent angle is equal to the critical angle. The
intercept times for two segments of the refractor for each
recording direction can be used to evaluate the throw of
the fault, or it can be computed with one of the other
interpretation techniques described in Section 5.6.

1
Slope = __
V2

Diffraction
1
Slope = __
V1

1
Slope = __
V1

5.4

Offset
S2

Depth

S1

Velocity V1

Velocity V2

Fig. 5-10. Raypaths and timedistance curves for two


layers separated by a faulted horizontal interface.

(1984). Several authors have proposed techniques to


solve the hidden layer situation. For example, Soske
(1959) suggested making additional shots at a deeper
depth so that the hidden layer is no longer hidden;
Morgan (1967) proposed the use of a combined reflection and refraction approach. Palmer (1980) indicated
that an evaluation of the XY distance in the generalized
reciprocal method (GRM) could give information about
the possible presence of a hidden layer (discussed further in Section 5.6.5). Other papers that review the hidden layer problem include Green (1962), Kaila and
Narain (1970), Banerjee and Gupta (1975), Greenhalgh
(1977), Raghava and Kumar (1979), Whiteley and
Greenhalgh (1979), and Pant and Raghava (1987).

5.3.7

Two Layers Separated by a Faulted


Interface

Figure 5-10 shows raypaths from source locations at


S1 and S2 to several receiver locations for two layers,
with the lower layer faulted vertically. Diffractions occur
when the source is on the upthrown side of the fault (S1);
these are observed as first arrivals for a range of offsets
before being overtaken at larger offsets by refracted
arrivals from the deeper segment of the interface. With
the source on the downthrown side of the fault (S2), the
refraction waves travel through the high-velocity layer
and do not emerge as wavefronts at the critical angle. At

DATA ACQUISITION OF
REFRACTION DATA

Refraction data can be acquired either by a separate


refraction crew in the field, often called a weathering or
low-velocity layer (LVL) crew, or by using the first
arrivals recorded as part of a seismic reflection survey.
The latter approach is now more appropriate than it was
a few years ago because the group interval, and hence
array lengths, are much smaller, thereby minimizing the
attenuation of the refracted arrival. A description of each
approach, including their advantages and disadvantages, follows in Sections 5.4.1 and 5.4.2.
The offset range is chosen so that refracted arrivals
are observed from the target refractor as first arrivals.
The group interval should be commensurate with the
detail required, and recording should be done in two
directions. If regular off-end recording is used for the
seismic reflection survey, the data can be gathered on a
common-receiver-location basis to simulate source locations at the other end of the spread, providing there is
adequate fold of cover.
The traditional method of acquiring refraction data is
with a point source and a single geophone, or bunched
geophones, because the main requirement is for an adequate signal-to-ambient noise ratio. The signal-to-coherent noise ratio is not an issue for refraction recording
because the refraction arrival arrives before the ground
roll and other noise modes. This is a completely different situation to that of a reflection survey.
To obtain good estimates of refraction arrival times,
the source and receiver should be as broadband as possible, minimal filtering should be applied to the data
recording (Hagedoorn, 1964), and the signal-to-noise
ratio must be adequate. These are the same basic
requirements described in Section 4.2 for the data acquisition parameters of uphole surveys. These specifications should be adhered to in LVL surveys but are
unlikely to be practical on a seismic reflection survey in
which the parameters are dictated by the requirements
of the reflection survey. In addition, timing checks must
be carried out on the whole recording system, from the
time-break through to the camera display, to ensure that
any delays are understood and estimated. These delays
are then applied as arrival-time corrections or adjustments (see Section 5.5.7). Relevant LVL data should be

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


archived for future reference (see Section 3.11). The
refraction arrivals on a seismic reflection survey form
part of the main data set, and these are routinely
archived.
Picking refraction arrivals and making necessary
time adjustments are covered in Section 5.5, and various
methods for interpreting refraction data are described in
Section 5.6. Velocities down to the refractor should ideally incorporate information from other techniques,
such as uphole surveys, to minimize depth conversion
problems related to possible thin layers and velocity
inversions.
A novel approach to marine refraction surveying was
proposed by Hunter and Pullan (1990) in which several
offset shots are recorded by a vertical array of
hydrophones suspended in the water. The technique is
intended for use in ice-covered areas where it is more
efficient than laying out a cable or deploying individual
receivers on the water bottom. It is shown that for horizontal layers, the refracted arrival time from layer n (tn)
can be given by

Vn2 V02
x
tn =
+ ( hr + hs )
VnV0
Vn
n 1

+ 2 zm

(V

2
n

Vm2

1/ 2

1/ 2

VnVm

m =1

(5.33)

where x is the source-to-receiver offset, hr and hs are the


receiver and source heights above the water bottom, zm
the thickness of layer m, and Vm, V0, and Vn the velocities of layer m, the water, and the refractor, respectively.
Thus, a plot of the first-arrival times against the
height of the receiver above the water bottom can be
used to compute the refractor velocity. The slope of a
segment of this plot (Sn) is given by

Sn

(V
=

2
n

V02
VnV0

1/ 2

(5.34)

If the refractors dip, or the receiver array is nonvertical, numerical modeling is normally required. Another
example of this technique is shown in Hunter et al.
(1991).
To analyze local subwater-bottom anomalies, such as
the low-velocity zones described in Section 2.4.2, Wilson
(1983) and Zachariadis (1986) patented the use of combined reflection and refraction techniques. Time delays
caused by the anomalous zone are estimated by comparing the information from normal and anomalous
areas. With a single ship operation, sources are located
both behind the vessel and on the water bottom;
receivers are also located on the water bottom.

5.4.1

153

Low-Velocity Layer (LVL) or


Weathering Surveys

Low-velocity layer (LVL) or weathering surveys are


generally conducted at discrete locations along a line
and not on a continuous basis. For simplicity, they
should be recorded on straight segments of the line
where the elevation change is minimal. If this can be
achieved, undue complications are not introduced into
the interpretation stage. The surveys are undertaken by
an LVL or weathering crew, which is normally a small
unit attached to the main seismic crew. Basic facilities
and logistical support are shared with the main crew,
but the crew operates independently along the survey
line and normally records data just before or soon after
the recording of the main survey data. In normal usage
the field technique is limited to fairly shallow targets
unless several additional shots are recorded into each
spread.
A typical LVL crew is equipped with a 24-channel
recorder and uses a group interval in the range of 520
m, often with a smaller interval for groups close to the
source location. The minimum offset is typically only a
few meters so that the direct arrival is recorded on the
near groups and is used to estimate the near-surface
velocity. The total spreadlength is thus typically in the
range of 100400 m, and it is normal practice to record
into the spread from both directions. Figure 5-11 shows
an example of a spread layout and low-gain field record
displays from a reversed profile. Because the group
intervals along the spread in this example are not equal,
the arrival times for a constant refractor velocity are not
linear on the field record.
In a simple two-layer case, Table 5-2 showed that
refracted energy occurs only as a first arrival at offsets
greater than several times the refractor depth. For example, if the refractor depth is 30 m, first arrivals are not
observed at offsets less than about 90 m when the refractor velocity is 2.5 times the near-surface velocity. In areas
where several near-surface layers are to be mapped, it is
likely that a large offset range will be required. If the
standard spread is of insufficient length to record the
deeper targets, the offset range can be increased with
additional shots that are offset from the recording
spread.
Figure 5-12 depicts a short reversed spread between
two source locations S3 and S4, together with two offset
shots S1 and S2 at one end of the spread. The first-arrival
times from the shot at S3 show that three are from the
direct arrival and nine from a refractor. The first-arrival
times from the source at S2 show that all twelve values
are from a refractor and that these parallel the times
from S3, referred to by Sjgren (1980) as the law of parallelism. If all first arrivals come from the same refractor,

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

154
(a)

Static Corrections for Seismic Reflection Surveys


SP A

SP B

Fig. 5-11. Recording spread layout and field records from an LVL or weathering survey: (a) recording spread layout;
(b) field record from SP A; (c) field record from SP B.

t2

Time

t1

S1

S2

S3

Recording spread

S4

Fig. 5-12. Timedistance curves to illustrate the use of offset shots to increase refraction coverage beneath a fixed
recording spread. Variables: t1 and t2 are shot-to-shot time constants between shots S2 and S3 and shots S1 and S2.
Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

155

150

Time (ms)

100

50

0
S-50

S-1

Recording spread
0

50

S1

S50

100 m

Fig. 5-13. Timedistance display from an LVL survey; shots located off both ends of a 115-m recording spread at offsets
of 1 and 50 m.

there is a constant time difference for each shot (composed of the differential time through the near-surface
layer at the two shot locations) and for the additional
distance along the refractor. If the differential time constant is estimated, shown as t1 in Figure 5-12, it can be
used to extend the arrival times from the short-offset
shot (S3) toward its source location. That is, refraction (or
pseudo-refraction) arrival times for the three short-offset
traces can be defined, even though one or more of these
traces may have an offset less than the critical distance.
The shot-to-shot constant t1 is estimated by averaging
individual time differences for all common receiver
locations, with the proviso that the times are from the
same refractor for both shots.
If the offset range is increased further, it is possible
that a deeper refractor will be mapped. This is shown for
arrivals from the source at location S1 in Figure 5-12, in
which the farthest four offsets have a faster apparent
velocity. They can also be identified as being on a different refractor by making shot-to-shot comparisons
between the S1 and S2 arrival times (t2 in Figure 5-12),
which indicate a decrease in this time difference with
offset.
In the field example shown in Figure 5-13, shots are
located at offsets of 1 and 50 m from each end of a 115-m
spread. There is considerable variation in the time differences between common receivers for different shots,
much of which is probably due to the quality of the first
arrivals because they do not have a sufficiently high sig-

nal-to-noise ratio. For the shots at S-50 and S-1, the differences for the deepest layer vary from 18 to 24.5 ms,
with an average of 21 ms. I discuss the interpretation of
this data set in Section 5.6.1.
The source used in an LVL survey can be dynamite
in a shallow hole, weight drop, land air gun, detonating
cord, or one of the other sources used in a shallow
reflection survey (listed in Section 3.8.5.2); these include
the hammer and plate technique for short-offset recording. For low-energy sources, the field recorder should
have the capability of summing records prior to the
production of the field monitor. Igarashi and Matumae
(1985) showed an improvement in the quality of the
first break with an increase in the number of field
records summed. I stated earlier that source and receiver arrays are not required, so that a single source unit
can be used and either a single geophone or bunched
geophones.
The interpretation of data recorded by a refraction
crew is usually performed in the field. This involves
picking the first breaks and making any necessary
arrival-time adjustments (see Section 5.5). One of several interpretation methods are then applied (see Section
5.6), including Hagedoorns plus-minus method
(Section 5.6.4), the ABC method (Section 5.6.3), the generalized reciprocal method (GRM) (Section 5.6.5), or the
simple intercept-time method (see Section 5.6.1) used on
the model data in Section 5.3. In areas where it is desirable that the source for production dynamite data be

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

156

Static Corrections for Seismic Reflection Surveys


located below the weathered layer, analysis of LVL data
can sometimes assist in defining the required hole depth
along the line.
Some advantages of using a special refraction (LVL)
crew for weathering control are as follows:
1. The point source and point receiver generally
give good quality first breaks.
2. Estimates can be made of the near-surface velocity from direct arrival information.
3. The weathering thickness can be estimated prior
to drilling deep holes on a dynamite survey.
Disadvantages of using an LVL crew for weathering
control include the following:
1. The additional field crew causes cost increases
and potentially more operational problems. Costs
increase even more if the spacing between survey
locations is reduced or if continuous coverage is
required.
2. A normal survey without extended offsets allows
only for a shallow refractor to be mapped.
Additional shots can be used to map slightly
deeper refractors, but at a further increase in cost.

5.4.2

Fig. 5-14. Field record from a seismic reflection crew


showing refracted arrivals; group interval of 25 m,
spreadlength of 2975 m.

Refraction Data Recorded on


Seismic Reflection Records

In many areas, the first coherent energy seen on a


seismic reflection record is associated with a refractor, or
in the case of near offsets, it is a direct arrival through
the near-surface layer. Section 5.3 showed that the
apparent velocity of the refraction arrival depends on
the refractor velocity and the dip of the refractor. In
more complex situations, there is also a dependence on
the dip and spatial velocity changes of other layers
between the surface and the refractor. In addition, it is
influenced by varying group intervals, differential
datum static corrections caused by elevation changes
and near-surface variations along the line, diffractions
(see Section 5.3.7), and a change of refractor. For
crooked-line recording and some 3-D surveys, contiguous receiver locations can correspond to offset changes
smaller than a group interval, which give rise to apparent group interval changes across the record.
In most cases, the wavelengths of the refraction
arrivals correspond to those normally treated as coherent noise on a seismic reflection survey, or are intermediate between noise and reflection signal wavelengths.
Figure 5-14 shows a field record with a spreadlength of
2975 m in which two refractors and a direct arrival or a
slower refractor are present. The apparent velocities are
about 1825, 2000, and 2650 m/s, and the arrivals have a

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


dominant frequency of about 18 Hz. This translates to
wavelengths in the range of 100150 m.
Prior to the use of multichannel processing techniques such as velocity filtering, it was normal to design
field arrays to remove as much coherent noise as possible prior to recording the field data. The availability of
multichannel processing techniques has now made it
possible to use a system approach to attenuate coherent
noise. This generally means that the short-wavelength
noise is attenuated in the field (with the arrays acting as
a spatial anti-alias filter) and the longer wavelength
components are attenuated in the data processing center
(see Section 6.6.1). This approach has been made possible by an increase in the number of recording channels
and an associated reduction in the group intervals.
In most cases, refraction data fall in the long-wavelength component so that attenuation by the field arrays
using the new acquisition approach is small. The attenuation can be further reduced for surface sources if the
data are not summed (stacked) in the field, or alternatively if the refraction arrivals are picked prior to the
summation process. The effects of arrays and intra-array
static corrections on the shape and time of the refracted
arrivals are discussed in Section 5.4.3.
An alternative approach for recording high-quality
refraction data involves the use of a special geophone in
the receiver array which is the only one active at the start
of recording. Immediately after the first arrival, this geophone makes the remainding geophones in the array
active to attenuate coherent noise for the reflection
recording (Thompson, 1963). This field approach is
appropriate if the arrays are long, although its use under
these conditions has not been extensive; it is not considered necessary with the shorter arrays used on many of
todays crews.
The range of offsets used in a seismic reflection survey is usually large so that refracted energy is often
observed as a first arrival from more than one refractor.
If a small group interval and minimum offset are used,
the direct arrival through the near-surface layer may be
recorded, but in many instances, there will be insufficient data to obtain a reasonable estimate of the nearsurface velocity.
Interpretation of the refraction data can be undertaken by the field crew or done in the processing center and
involves picking the first breaks or first arrivals and
making any necessary arrival-time adjustments
(described in Section 5.5), followed by using one of several interpretation methods described in Section 5.6.
Except when the refractor is observed on a limited offset
range, a large amount of redundancy is normally available, allowing for a considerable amount of averaging or
high-grading to be performed. Various interpretation
methods are used, such as the generalized reciprocal

157

method (GRM) (Section 5.6.5), delay time method


(Section 5.6.6), or inversion method (Section 5.6.9).
Some advantages of using production seismic reflection records for an analysis of refraction data are as
follows:
1. There are no additional acquisition costs.
2. A large amount of redundancy is normally
available.
3. The source and receiver locations are those used
in the reflection survey.
4. Refractors can usually be mapped well below the
weathered layer and on a continuous basis.
Disadvantages of using production seismic reflection
records include the following:
1. The group interval and distance to the near
group may not allow for the mapping of relatively thin shallow layers or the direct arrival.
2. Some degradation of the first arrival may occur
due to the source and receiver arrays used, but
this is likely to be minimal if a small group interval is used.

5.4.3

Factors Affecting Shape and


Amplitude of Refraction Arrivals

Almost all published work on refraction data concentrates on the time of the arrival, and only a few examine
its amplitude and whether amplitudes can be used to
extract additional information from the data. The head
wave pulse approximates an integration of the direct
wave (e.g., Dix, 1961), and its dominant frequency is
inversely proportional to the cube root of the charge size
(OBrien, 1967b). The shape of the first arrival is influenced by ground conditions and coupling at both source
and receiver locations, and its amplitude varies with offset. The type of source and receiver used, the array
dimensions, and the magnitude of the intra-array static
corrections also have an effect. In addition, the shape of
the first arrival from a vibroseis survey is different from
one from a dynamite survey. This is because the data
include a component of the autocorrelation function of
the sweep signal and the recording filter after the sweep
removal process. This means that a causal wavelet, such
as one obtained from an explosive source, is not associated with correlated vibroseis data.
For long offsets, where the distance traveled along
the refractor exceeds several wavelengths, the head
wave amplitude decreases by a factor close to the
inverse square of the offset (e.g., Muskat, 1933; Heelan,
1953; OBrien, 1955, 1957, 1967a; Kumar and Raghava,
1981). At smaller offsets, which is likely to be the case for

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

158

Static Corrections for Seismic Reflection Surveys


Offset

(a)
(b)

Time

Amplitude

(c)
(d)
(e)
(f)
(g)

40

20

60

Time (ms)
Array

Fig. 5-15. Refraction arrivals showing shinglingen echelon pattern often associated with thin layers.

many LVL surveys, the amplitude decrease with offset is


slightly less (e.g., Levin and Ingram, 1962; Donato, 1964,
1965; Hatherly, 1982a).
In addition to the variation of amplitude with offset,
significant attenuation occurs if the refractor is thin
(Press et al., 1954). This effect is normally associated
with shingling, a phenomenon characterized by a shift of
energy to successively later cycles as the offset increases,
producing an en echelon pattern (Figure 5-15) which is
observed in some areas. Model experiments by
Lavergne (1961, 1966) indicated reasonable recording of
refraction arrivals for thicknesses greater than /6,
where is the dominant wavelength of the data. Poley
and Nooteboom (1966) reported a rapid increase in
attenuation when the thickness of a high-velocity layer
is less than 0.3. The observed velocities were reported
to be less than those of a thick layer. Theoretical aspects
have been discussed by Spencer (1965) and other data
examples shown by Cassinis and Borgonovi (1966) and
Trostle (1967).
When considering the effect of arrays, it is important
to remember what is being recordeda first arrival and
the wavelet associated with it. If an in-line receiver array
consists of six receivers or elements, the six first arrivals
occur at different times and the output from the array
attenuates some of the refracted energy. However, for
the closest receiver to the source, the first arrival is
summed only with the ambient noise recorded on the
other five receivers; thus, in theory, the first break seen
on the output should be very close to the output from
the nearest receiver.
This is illustrated in Figure 5-16, which shows the

Array

(a)
(b)

(e)

(c)

(f)

(d)

(g)

Fig. 5-16. Influence of the receiver (or source) array on


the refracted arrival: (a) point receiver (model wavelet);
(b) 6-element array; (c) 9-element array; (d) 12-element
array; (e) 6-element array; (f) 9-element array; (g) 12-element array (Note: time scale is the same for the arrays
and the refracted arrivals).

output from six different arrays, which are normalized


for display purposes. In this example, the output amplitude from each geophone in the array is constant. The
refraction wavelet in Figure 5-16a is based on the one
shown in Laster et al. (1967); it has a period of about 32
ms, which is a frequency of about 31 Hz. The six arrays
(Figures 5-16bg) have time-equivalent lengths of
522 ms, which equate to array lengths of 1044 m for a
refractor velocity of 2000 m/s, or 1566 m for a refractor
velocity of 3000 m/s. The first-break times are unaltered
by the array, but the troughs and peaks are delayed by
up to 14 ms, close to half the time-equivalent length of
the array. Table 5-4 lists the normalization factor, which
is almost equal to the attenuation by the array, the timeequivalent length of the array, and the times of various
troughs, peaks, and zero crossings. In most cases, this
shows that the times are delayed by about half the timeequivalent length of the array.
In practice, the shape and amplitude of the array output is influenced by all the receivers within the array
and is subject to variable ground coupling and intraarray static corrections. Thus, the overall effect of the
arrays can be expected to vary along the line, implying

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

159

Table 5-4. Array Length, Attenuation, and Picked Times of Wavelet Data in Figure 5-16.

Array

Array
Length
(ms)

Attenuation
(dB)

Picked Times (ms)

a
b
c
d
e
f
g

0
5
8
11
10
16
22

0
0.6
1.4
2.6
2.6
5.8
10.2

Trough

Zero
Crossing

Peak

Zero
Crossing

Trough

12
14
16
16
16
17
17

17
20
21
22
22
23
23

25
29
30
31
31
33
33

33
36
37
38
38
41
44

40
43
44
46
46
49
54

that uncertainty exists about the correct distance to use


for a specific receiver group. If good first breaks are
observed and used in the interpretation, then the distance should be the minimum distance between the
source and receiver arrays. If a later part of the wavelet
is used, then a more realistic distance is the regular array
center to array center distance.
The influence of intra-array static corrections is illustrated in Figure 5-17, where different directions of
recording are shown to produce different effects on the
data. Figure 5-17a shows uphill recording with a large
time difference across the array, while Figure 5-17b
shows downhill recording with only minor time differences because the refracted wavefront is nearly parallel
to the surface. In this example, the time shifts are well
organized, but a random or high-frequency component
is usually present as well. The intra-array static corrections are a function of the elevation profile (or more correctly, the relative static corrections across the array) and
the refractor and near-surface velocities, as these determine the angle of the emerging or refracted wavefronts.
These static corrections attenuate the first arrival and
time shift the output trace. Sometimes, a receiver other
than the one closest to the source is responsible for the
first break seen on the output trace.
The impact of intra-array static corrections on the
output data has been shown by Berni and Roever (1989)
and Steeples et al. (1990). In some cases, the resulting
signal distortion can be removed, or partially removed,
with a subsequent dephasing operator. Ongkiehong and
Askin (1988) showed the effect of random perturbations
on individual receivers on the array-formed outputs, in
both time and amplitude (see Figure 6-51). They demonstrated that well-defined noise was spread across the f-k
plane by these perturbations, indicating that intra-array
static corrections impacted coherent noise as well as signal. The impact of time shifts on multichannel processes
for coherent noise removal and multiple attenuation is
described in Section 6.6.1.
The quality of refracted arrivals is considerably better
than that of reflected data in most areas, such that refraction arrivals are often used to QC field data. Quality

control includes identification of reversed groups,


anomalous receiver groups (potentially due to changes
in local ground conditions), and defective receivers in
the array. In addition, the refraction arrivals can be used
to verify that the recording geometry matches the
planned arrangement, which is commonly done in 3-D
surveys. If an anomalous group can be recognized in the
field, there is an opportunity for the relevant group to be
investigated and any necessary remedial action taken.
Various types of displays that can be used to assist in the
QC of refraction data are covered in Section 5.5.6.

5.5

REFRACTION ARRIVAL PICKING


AND ARRIVAL-TIME
ADJUSTMENTS

The first step in any refraction interpretation,


whether a survey for near-surface control or for deep
targets, is to review and pick the data. The review phase
is concerned with an analysis of the data to be picked,
such as the offset range or ranges to be used in the interpretation, which depends on the target refractor depth.
In addition, some interpretation techniques require an
initial estimate of the refractor velocity. This analysis is
often undertaken using conventional field record displays with sufficient gain to see the first arrivals. In the
case of crooked-line or 3-D recording, or in the presence
of large near-surface changes, the apparent refractor
velocity can vary significantly along the recording
spread and, in some cases, can even indicate a negative
velocity. It is therefore important for care to be taken in
estimating the refractor velocity.
The picking is often performed on a set of reduced
traveltime displays, that is, after removal of the refractor
velocity component, a procedure also referred to as
refractor stepout or moveout (equal to the trace offset
divided by the approximate target refractor velocity).
Alternative displays are described in Section 5.5.6,
including the offset panel (Fulton and Darr, 1984, 1985),
constant-offset display, and gated source and receiver

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

160

Static Corrections for Seismic Reflection Surveys

(a)
Array

Wavefront

Refractor

(b)
Array

Wavefront

Refractor

Fig. 5-17. Effect of intra-array static corrections on the


refracted arrival: (a) recording uphill; (b) recording
downhill.

displays, in which small segments of data are displayed


in a compact and informative format. These types are
mainly used for QC, but can also be used to define
search windows or gates for a subsequent automated
picking technique.
In some cases, a limited amount of processing can
also be performed, such as scaling and filtering, as a preconditioning step. Care must be taken to ensure that any
precursors introduced by the processing, such as the
side lobes of a digital filter, are sufficiently low in amplitude not to introduce a false first arrival. Additional preconditioning may be required for vibroseis recording,
especially if an attempt is made to pick a first break (see
Section 5.5.4).
I indicated in Section 5.1 that source and receiver
stacks could also be used in the interpretation of refraction data (Propes, 1991; Ostrander, 1992). Potential
advantages are that the signal-to-noise ratio is
improved, the data can be displayed as a time section,
and the number of traces to be picked is significantly
reduced. An optimum stack requires the correct refractor velocity for data reduction (refractor stepout or
moveout) and minimal time shifts within the data to be
stacked. Time shifts can be due to structural variations,
variations from receiver to receiver for the source stacks,
and from source to source for the receiver stacks. The
concept of source and receiver stacks is discussed fur-

ther in Section 7.3.2, where it is used to estimate residual


static corrections; optimum results often require an iterative approach.
A refraction profile along the line can be obtained by
picking the final common-surface stacks; this normally
contains predominantly long-wavelength (low-frequency) information. The short wavelengths are obtained
from the time shifts necessary to optimize the stacked
sections, which are equivalent to residual static corrections for the common-surface-stack method described in
Section 7.3.2. Because the stacks are performed on
reduced data, with the refractor stepout removed, the
time profile is equivalent to a delay time profile (see
Section 5.6.2).
A common-midpoint (CMP) stack, using the same
traces that are used in a reflection CMP stack, has also
been used (e.g., Coruh et al., 1993). For optimum results,
this requires that differential source and receiver delay
times (see Section 5.6.2) are evaluated and applied,
together with the removal of the refractor stepout; this
procedure should minimize the smearing effect of the
near-surface layer. The data summed by this approach
tend to smooth out some of the refractor structure, as
different offsets traverse different parts of the structure.
Refraction arrivals can either be picked manually or
by an automatic technique, which normally includes a
manual or interactive editing capability (see Sections
5.5.1 and 5.5.2). The aim of picking the first arrival is to
obtain a good estimate of the refraction arrival time for
each trace, which is then used with one of the interpretation methods described in Section 5.6 to derive a nearsurface model for the data. In most situations, the picking accuracy requirements are not as severe as those
required for uphole surveys. However, apart from poor
quality data, picks should be timed so that the timing
error does not exceed 1 ms, or 2 ms in some areas. In the
case of high-resolution surveys, it is often necessary to
improve on these specifications.
Specific problems associated with picking second
arrivals, required in some areas, are discussed in Section
5.5.3. Instead of picking individual traces, pairs of traces
can be crosscorrelated to produce a relative time profile
along the line (see Section 5.5.5). Various displays are
used for QC of the picked data, as shown in Section
5.5.6. Refraction arrival-time adjustments, which may
need to be applied to the arrival times, are documented
in Section 5.5.7.

5.5.1

Manual Picking of Refraction


Arrivals

There are many similarities between picking refraction arrival times and uphole times (see Section 4.4.1),
including requirements of a broadband signal, minimal
filtering of data, a good signal-to-noise ratio, and a high-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

t2

Amplitude

t1

t4
t6

t0

t5

t3

t7

Time

Fig. 5-18. Refraction arrival to illustrate refraction picking options. Variables: t0 is time of first break, t1 is first
arrival time through first inflection point, and t2 to t7 are
trough, zero crossing, and peak times (after Laster et al.,
1967).

gain display. In Section 4.4.1, I quoted from Ricker (1953,


20) (see also Ricker, 1977, 149): There is no sudden takeoff of the trace when the disturbance arrives. The motion
begins gradually and if a first kick arrival time is
attempted, the time picked will depend upon the overall magnification of the seismograph. This was illustrated with examples in Figure 4-10. Ricker suggested
using the intercept on the time axis of the tangent
through the first inflection point; this is slightly later
than the true first-arrival time, but it is less sensitive to
the signal-to-noise ratio and overall gain. In a description of an automatic picking technique, Hatherly (1982b)
further developed this inflection point concept.
Figure 5-18 shows a refraction arrival in which the
first break is indicated by time t0, with other times noted
for troughs, zero crossing points, and a peak. On very
high-gain displays, the first break or kick can be
extremely sharp, approaching 90. Using the tangent
through the first inflection point (Ricker, 1953) results in
the estimated first-arrival time at t1. In common with the
picking of uphole times (Section 4.4.1), later phases of
the refraction arrival can be picked, such as a peak,
trough, or zero crossing, as shown by times t2 through t7
in Figure 5-18.
Interpretation techniques require either first-break
times or, in some cases, a relative time profile that is subsequently converted to absolute depth with external control, such as from uphole survey locations. To convert to
first-break times, a simplistic approach assumes that there
is a constant time shift between the first break and the
peak, trough, or zero crossing picked. Thus, it is assumed

161

that no change of frequency occurs with offset, receiver,


or source location, which is a reasonable assumption for
many marine surveys but not for land surveys. This was
highlighted by examples in Figures 4-11 and 4-12. In some
cases, however, this approach can lead to a smaller error
than trying to pick poor quality first arrivals.
The frequency changes with offset and location can
be monitored by estimating, for example, the times of a
consecutive peak and trough. This information can then
be used to improve the estimated first-break time from
the trough, peak, or zero crossing time picked. Even
when absolute first-break times are not required by the
interpretation technique, it is still assumed that the time
between the relative time profile and the individual
first-break times is constant. If this is not the case, an
error is introduced into the interpreted time and subsequent depth profile.
Accuracy requirements are dependent on the survey
objectives, the field parameters, and the velocities and
thicknesses of near-surface layers. As indicated earlier,
an accuracy of 1 or 2 ms for individual picks should be
the goal. In some cases, averaging is done to identify
wild or spurious picks and to reduce any random picking errors.
The seismologists or interpreters experience and
ability to perform pattern recognition on poor quality
data can be a valuable asset in picking refraction arrivals
manually. Manual picking is a practical approach when
working with LVL surveys, but the volume of data
acquired on most seismic reflection surveys makes it
impractical unless a small subset of the data is picked.
Manual picking was once standard when the data
acquired each day could be picked and timed by a seismologist in a few hours, but this ceased to be the case
once the number of recording channels exceeded 24 or
48. The manual method can be speeded up if a digitizer
is usedwith the seismologist either picking and timing
the refraction arrivals directly on the digitizer or using
the digitizer to time the refraction arrivals.

5.5.2

Automated Methods

The need for refraction arrivals to be picked with a


computer technique arises when the amount of data
acquired can no longer be handled manually, even if a
digitizer is used to speed up the process. This is usually
the case when the data are recorded as part of a seismic
reflection survey, even if only a subset of the full data set
is picked because the refractor to be mapped is observed
only over a specific offset range. A large number of
approaches can be used to pick refraction first arrivals,
each with its own advantages and inherent limitations.
In most cases, they work well when the signal-to-noise
ratio is high and the first arrival is clearly defined.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Static Corrections for Seismic Reflection Surveys

Ideally, they should also be capable of performing well


when the recorded data are not up to this standard,
although clearly not for the extreme case of NG data.
However, the surface redundancy typically available
means that some poor estimates can be tolerated, as they
can be excluded later once comparisons are made with
other picked times.
The signal-to-noise ratio of the first arrival can be
improved by stacking several consecutive traces together
(e.g., Ervin et al., 1983; Gelchinsky and Shtivelman, 1983)
or by applying a coherency filter (e.g., Coppens, 1985).
This assumes that the refractor moveout (stepout) is
taken into account as part of the process, or is removed
from the data, and that no large trace-to-trace variations
exist that would significantly attenuate the signal as well
as the noise. Refraction arrivals are normally picked on
source-ordered data, but data organized on a receiver or
offset basis are also used (e.g., Coppens, 1985). In noisy
areas, or where there are large changes in arrival time
from location to location, the ability to examine the data
interactively in more than one domain can be very useful.
Techniques used range from a simple examination of
the values down the recorded trace to those based on the
power of the trace, its envelope or other attribute, and
how these change with time. Another automated
method that uses crosscorrelations is described separately in Section 5.5.5. In many automated schemes, the
picking is done in a batch mode, that is, all data are picked
without manual intervention. In contrast, others have an
interactive element allowing the user to play a part in
the picking process by changing relevant parameters
along the line. In many cases, the preferred pick time is
stored with one or more possible picked times on either
side of it which can be used instead of the preferred pick
as a result of an analysis of other picked times in a pick
validation scheme. In this section, I describe the different picking schemes and conclude with a few comments
on pick validation and editing.
The simple approach for automatically picking the
first break involves scanning down the trace for the first
value that exceeds a preset threshold; this can be adapted to search for and time the first peak or trough. The
disadvantage of this approach is that if the threshold is
too low, spurious picks due to noise are obtained, and if
it is too high, genuine events or first breaks may be
missed. If the trace is convolved with a box car or similar filter, the first break is transformed into a peak, generally slightly easier to pick than the first break (Ervin et
al., 1983; Gelius et al., 1984; Ramananantoandro and
Bernitsas, 1987). Khan (1994) recommended conversion
of the recorded data into square waves for subsequent
picking.
The amount of freedom to search the field data can be
reduced with appropriate parameters, such as the spec-

Maximum

Energy

162

Threshold

t3

t2

t1

Time

Fig. 5-19. Energy variation down a trace to illustrate


picking of a refraction arrival. Variables: t1 to t3 are estimates of refraction arrival times (see text for details).

ification of a window containing the first arrival, a


process that can be defined by input parameters or
described interactively on a work station. In this situation, some of the mispicks referred to earlier are avoided, and other possible picking schemes can be tested
and used, including the largest peak in the specified
window, the first peak, or the peak closest to the expected arrival time based on information from previous
sources and receivers.
Section 5.5.1 pointed out that care must be exercised
with methods that aim to pick a later phase of the first
arrival rather than the first break. In automated methods, it is a relatively simple task to monitor changes in
frequency with source and receiver location and offset,
which can be used to extrapolate back to a first-break
time. This requires calibration on some traces in which
times for both the first break and the phase picked are
available. The time adjustment associated with changes
in observed frequency can be averaged for each source
and receiver location because these factors are often
larger than the offset-related effect. As an example,
Hatherly (1982b) computed the first-break times from
the first inflection point picked by adjusting them with
the average difference from the first break to the inflection point times for each field record.
Other picking schemes involve analyses of the energy variation down the trace, in which the energy is computed by squaring the value at each sample. These
schemes include analysis of the energy peak, differences
between samples, average energy variations over a window, and variations in the energy envelope. As with
other power schemes, energy variations are dominated
by the noise component rather than signal amplitude
variations once the signal-to-noise ratio drops below 1.
Figure 5-19 shows schematically the variation of energy

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


with time down a seismic trace containing a refracted
arrival. One approach is to set a threshold at a percentage of the maximum value, corresponding to time t2 in
Figure 5-19, and use this as a first estimate of the refraction arrival time. The picked time can then be modified
by a fraction of the waveforms dominant period or the
original trace can be searched for a peak or trough close
to the estimated pick time. Alternatively, the slope near
the threshold can be extended back to the axis to give
another estimate of the pick time, as shown by time t3 in
Figure 5-19. If high-amplitude signal is present immediately after the refraction arrival, the above techniques
can indicate too large a value for the initial estimate of
the refraction arrival time.
Seismic attributes and their ratios (Taner et al., 1979)
can also be analyzed to give information on refraction
first-arrival times. Various techniques for picking refraction arrivals are documented in the literature, several of
which are summarized below, with others relating to
vibroseis data or the crosscorrelation approach (Sections
5.5.4 and 5.5.5).
Hatherly (1980, 1982b) estimated the first data on a
trace which was statistically different from the prerefraction arrival data. A crossplot is generated of each
sample on a trace against the following one. It shows a
fairly narrow distribution of values (trace amplitudes)
about a linear slope for the prerefraction arrival noise
until the seismic signal arrives, characterized by several
consecutive values outside the normal distribution limits defined at earlier record times. These events are then
checked by a linear least-squares prediction approach,
in which a linear operator is used to predict values from
earlier data. The first arrival is identified by a large prediction error because it is less predictable than the prerefraction arrival noise. The trace is then searched locally for the first inflection point and first break, and an
average value for the difference between these two
times is then applied to all inflection point times.
Sims and Mackenzie (1973) computed the average
absolute amplitude ratios of successive wavelets, which
were defined as the times between zero crossings. The
first ratio above a preset threshold indicates the approximate location of the first arrival, which is more specifically given by the time of the peak within the selected
wavelet. They found that two modifications increase the
success rate of the technique: starting the analysis after
the first arrival and working backward toward time
zero, and computing the amplitude ratio over several
consecutive wavelets rather than just one.
Chun and Jacewitz (1980, 1981, 1983) have described
an approach in which all possible peaks and troughs in
a window centered on the expected time of the first
arrival were picked (Figure 5-20). The initial estimate of
the refraction arrival time corresponds to the time when

163

the power ratio between successive time picks is at a


maximum. Ketelsen and Fromm (1982) also examined
the growth of the signal by making comparisons using a
window moving down the trace.
Coppens (1985) computed the energy ratio of a short
gate, which was nearly equal to the period of the first
arrival, versus a longer gate, which ranged from zero
time to the end of the short gate, for a range of positions
of the short gate. The time of the peak in the energy
ratio gives an estimate of the first-arrival time, and the
seismic trace is then searched for the nearest peak or
trough. Several examples of this technique are shown in
Figure 5-21.
Farrell et al. (1981) computed the ratio of the reflection strength of a window following a given sample to a
window leading the same sample. This ratio is at a maximum when the leading window is over the prearrival
noise and the window following the sample encompasses the first arrival. Reflection strength is one of the
seismic attributes defined by Taner et al. (1979) in their
paper on complex seismic trace analysis.
The analysis of several seismic attributes of all peaks
or troughs in a time gate was proposed as a supervised
learning technique by Taner et al. (1988). Attributes
included phase, amplitude, slope of the envelope, and
the power ratio (average power in a gate prior to a sample divided by the average power following the sample). Several crossplots, such as amplitude versus power
ratio, are generated and analyzed to find those which,
with suitable delimiters, give the most similar results to
hand picks on a sample of the data. This requires that
the chosen crossplot shows data clustered together and
not scattered over the whole plot. The selected ratios
with their delimiters, which in effect define a multidimensional space, are then used to pick the data; a similar approach was reported by Kusuma and Brown
(1992).
Similar attributes are used in a neural network
method described by Wagner et al. (1990). The network
is adjusted to ensure that its picked times are similar to
those obtained with a manual approach. An extension
from a single to a multitrace analysis was proposed by
Fuller and Kusuma (1993) and Kusuma and Fish (1993).
Pattern recognition and other neural methods were
described by Veezhinathan et al. (1991), Mural and
Rudman (1992), McCormack et al. (1993), and McCormack and Rock (1993).
In any automated picking process, validation checks
can be made to minimize the number of mispicks. For
example, a simple comparison of the refraction arrival
times from one source to two consecutive receiver locations can be made to verify that the time difference is
within an acceptable limit, after taking the refractor
moveout into account. This principle can be extended so

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

164

Static Corrections for Seismic Reflection Surveys

(a)
0.0

6
7
7

6
7
2

6
6
7

6
6
2

6
5
7

6
5
2

6
4
7

6
4
2

6
3
7

6
3
2

6
2
6

6
2
1

6
1
6

6
1
1

6
0
6

6
0
1

5
9
6

5
9
1

5
8
6

5
8
1
0.0

0.1

0.1

0.2

0.2

0.3

0.3

0.4

0.4

0.5

0.5

0.0

0.0

0.1

0.1

0.2

0.2

0.3

0.3

0.4

0.4

0.5

0.5

(b)

Fig. 5-20. Refraction arrival picking using a peak-trough ratio technique: (a) seismic data after removal of the refractor
moveout or stepout; (b) display of all peaks and troughs, with the maximum ratio between successive events (the preferred pick) indicated by a thicker line (Chun and Jacewitz, 1981; preprint).

that appropriate picked times from consecutive traces


are joined together to form a refractor segment or event
(e.g., Paulson and Merdler, 1968; Chun and Jacewitz,
1980, 1981, 1983). In forming or extending a refractor
segment, the preferred pick may be outside the acceptable differential time limit; if an alternative pick has
been made, it should be tested to see if it can be used to
extend the refractor segment. In some situations, cycle
skips or leg jumps will occur in the refractor segment,
especially if the maximum allowable differential time
shift is large. If all refraction arrival times for a source
location can be summarized by a few long segments,
this indicates a consistent set of refraction arrival times.
An alternative verification approach was described by
Coppens (1985) that involved a prediction error filter
used on the time profile to indicate when it was necessary for a more appropriate value to be selected as the
first arrival.
These approaches allow for some wild picks to be
identified and excluded prior to interpretation of the
data. With CMP recording and its inherent surface
redundancy, comparisons can also be made from one

source location to the next to ensure that the differences


between a pair of receiver locations are similar. These
comparisons are often made during the interpretation
stage, that is, after the picking phase, when average or
median values can be computed. Validation checks can
be performed in the batch mode or interactively at a
work station.
Once the arrival times have been picked, various QC
procedures can be implemented. This may involve a display of the picked arrival times or of the data, showing
the alignment of the refracted arrival after the picked
times have been removed or subtracted from the data,
or the picks superimposed on the refraction data
(Section 5.5.6).

5.5.3

Second Arrivals

The emphasis in this chapter is on the use of refraction first arrivals, or first breaks, to gain information
about the near surface; the topic of second arrivals was
briefly described in Sections 5.3.1 and 5.3.6. By definition, they arrive later in time than the first arrival. In

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

165

Fig. 5-21. Picking of refraction first arrivals with a moving


energy window technique; input seismic trace S(t) and
energy ratio trace F(): (a) good quality refraction arrival;
(b) noisy data; (c) vibroseis data. Variables: W1 and W2 are
gate lengths and 0 is picked time (after Coppens, 1985).

their simplest form, they are an extension of the direct


arrival at offsets greater than the crossover distance, or
the refracted arrival at offsets between the critical and
crossover distances (see Figure 5-4). Examples of using
second and later arrivals in refraction work have been
provided by Gamburtsev (1946) and specifically for
near-surface surveys by Naik et al. (1980).
Second arrivals can be used to identify the presence
of a hidden layer, as indicated in Section 5.3.6. The
example in Figure 5-8 showed that if the second arrival
(with velocity V2) was observed, an additional layer
could be defined. In some cases, however, the velocity
estimated from second arrivals may represent the veloc-

ity of a converted wave from one of the refractors picked


as a first arrival.
At offsets close to the crossover distance, second
arrivals occur at the same time as later phases of the
first-arrival pulse or wavelet. Under these conditions,
the second arrival can sometimes be observed on a lowgain display. At offsets farther away from the crossover
distance, the time separation between the first and second arrivals increases, leading to slightly easier identification of the second arrival.
If the second arrival is timed, a trough, peak, or zero
crossing is normally used because it is seldom possible
to time the onset of the pulse. These times can be used to

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

166

Static Corrections for Seismic Reflection Surveys

generate a relative time profile along the line which can


be converted to an absolute depth profile with external
control, such as from uphole survey locations. However,
as indicated in Section 5.5.1, this assumes that the time
between the time picked and the first-break time is constant. I suggested that the simplest way to compensate
for changes in frequency from trace to trace was to estimate the period of the data by picking a consecutive
peak and trough.

5.5.4

Additional Requirements for


Vibroseis Data

For data recorded with vibroseis as a source, a first


break is not obtained with conventional processing. I
mentioned in Section 5.4.3 that vibroseis data are normally two-sided after sweep removal because a component of the autocorrelation function of the sweep signal
and recording filter are included, such that the first
arrival contains precursors or prearrival side lobes.
Most vibroseis surveys require that data from several
source locations are summed and the data correlated.
The detrimental effect of array length on the refraction
arrivals can be minimized by not summing the data in
the field or by picking the first-arrival times prior to
source point summation, as suggested in Section 5.4.3. If
intra-array static corrections are significant in the area,
then the nonsumming approach has a further potential
benefit for the reflection data because relative datum
static corrections can be applied prior to source point
summation. However, the in-line dimensions of the
array for an individual sweep can be considerable and
depend on the number of vibroseis units used and their
spacing. If this array or subarray length attenuates the
refraction arrival significantly, dual recording can be
used in which a single vibroseis unit acquires the refraction information. Providing appropriate sweep parameters are used, single unit data can be acquired simultaneously with the main seismic reflection data (e.g.,
Garotta, 1983, 1984, 1987; Pritchett, 1991).
The above points refer to vibroseis recording in a
seismic reflection survey; it is rarely used in a weathering or LVL survey. However, some LVL data are
acquired using sources such as Mini-Sosie (tradename
of SNPA) (Barbier et al., 1976) in which the correlation
process may produce prearrival side lobes.
Various approaches are documented on the best way
to pick correlated vibroseis data. Ketelsen et al. (1983)
and Brtz et al. (1987) suggested the use of a two-sided
recursive shaping filter to convert the vibroseis
processed data to minimum phase, as shown in Figure
5-22. After this procedure, the data are then picked in the
same way as conventional data. Alternative techniques
to convert data to minimum phase include the decon-

Fig. 5-22. Preconditioning of vibroseis data with a twosided recursive shaping filter designed to give a minimum phase output; dynamite data on trace 1, vibroseis
data on trace 2: (a) field data; (b) minimum phase data
(Brtz et al., 1987).

volution-based approaches described by Ristow and


Jurczyk (1975) and Gibson and Larner (1984).
Farrell and Euwema (1984a) recommended that the
dominant peak of the arrival on conventionally
processed vibroseis data should be picked because it
corresponds to the zero-phase Klauder wavelet, which
is the sweep autocorrelation wavelet of the vibroseis
source (Hoover et al., 1984). A pragmatic approach was
indicated by Coppens (1985) who recommended that a
constant phase of the data should be picked along the
line to give a relative time profile (see Figure 5-21c). An
adaptive picking approach was described by Spagnolini
(1991) involving a wavelet recognition technique based
on a complex matched filter that was used to improve
the accuracy of the picked times.
Crosscorrelation techniques (see Section 5.5.5) are
applicable to vibroseis and nonvibroseis surveys. A relative time profile is generated that can be converted to

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

G(t)

H(t)

t1

t2

GH()
Negative lags

0 lag

Positive lags

Fig. 5-23. Crosscorrelation function GH() between two


input traces, G(t) and H(t). Variables: t1 and t2 are data
(gate) start and end times and is lag time.

depth using additional control, such as uphole survey


information. This approach assumes, however, that
there is a constant time shift between the first break and
the phase of the data (trough, peak, or zero crossing)
picked. That is, no change of frequency with offset,
receiver, or source location can be present. Possible
errors can be reduced if the frequency of the data is estimated and used to adjust the individual times (see
Section 5.5.1).

5.5.5

Crosscorrelation Techniques

A crosscorrelation approach can be used to obtain


estimates of differences in refraction arrival times. The
approach cannot produce absolute arrival-time information, obtained when first-break times are picked on a
refraction survey. Thus, a crosscorrelation approach
requires additional information to calibrate the data,
such as from an uphole survey or first-break times on a
few profiles, to produce a final depth model from the
refraction data.
Figure 5-23 shows schematically two input traces and
the resulting crosscorrelation function. The crosscorrelation peak occurs at a negative time, indicating that a
time shift occurs between the two traces. The input
traces should have an adequate time-bandwidth product and be similar in character for the crosscorrelations
to have a reasonable peak to side lobe level. This is
required for a clearly defined dominant peak, which can
then be picked to yield an estimate of the time difference
between the two traces. Variations in pulse shape or
character of the refracted arrival, however, can often be

167

tolerated if a suite of crosscorrelations are summed prior


to picking, a technique described later. Peraldi and
Clement (1972) suggested the use of a model trace,
formed by hand aligning two input traces, to improve
the quality of the crosscorrelations.
If the gate length used is too short, the crosscorrelations are unlikely to be of good quality; if it is too long,
reflection and refraction energy and coherent noise will
be included and produce suboptimum crosscorrelations
(Peraldi, 1969; Sims and Mackenzie, 1973). The crosscorrelation technique compares the time of the refraction
arrival and not just the first break. Therefore, changes in
near-surface conditions that result in different frequency
characteristics from successive source or receiver locations will give slightly different differential times from
those obtained by analysis of first-break times.
Crosscorrelations are used extensively in the evaluation
of residual static corrections and are described further
in Section 7.2.2.
The normalized crosscorrelation function GH()
between two input traces G(t) and H(t) is given by
t2

GH ( ) =

G(t)H(t + )dt

t = t1

t2
t2

G 2 (t)dt
H 2 (t)dt

t = t1
t = t 1

1/ 2

(5.35)

where is the lag time and t1 and t2 are the data (gate)
start and end times. The trace data used in the crosscorrelation should not be limited to the data start and end
times because this causes some of the multiplications for
nonzero lag times to be of live data times zeros. This
results in a subsequent reduction of the crosscorrelation
amplitude, a factor that can be appreciable with short
time gates and large lag times. Thus, the length of the
input trace used should be at least the gate time plus the
maximum lag time used in the crosscorrelation.
The crosscorrelation function GH() is scanned for its
maximum peak value to give an estimate of the differential time (or time shift) between the two input traces.
The time should be estimated to the nearest millisecond,
either by interpolation using values on either side of the
largest value in the peak or by resampling the data to a
fine sample period and picking the time of the largest
value. The normalized crosscorrelation has an amplitude of 1.0 if the two traces are identical. As the amplitude decreases, there is an increase in the uncertainty
about the time shift between the two traces. Below a specific threshold, which corresponds to a poor quality
crosscorrelation, it is generally preferable not to pick a
time shift.
In many cases, especially with poor quality data,

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

168

Static Corrections for Seismic Reflection Surveys


R3
R2

R1

V1

V2

Fig. 5-24. Raypaths to illustrate crosscorrelation of


several receivers from a common source location.
Variables: V1 is near-surface or weathered layer velocity
and V2 is refractor velocity.

there is no dominant peak and it is possible to pick several potential peaks. The conservative approach to this
situation is to pick the one closest to zero time or the
expected differential time. As in automated picking
(Section 5.5.2), alternative picks should be made and
subsequent validity checks used to decide on the most
appropriate one.
If one of the two input traces is reversed in polarity, a
dominant trough rather than a peak should be observed
on the crosscorrelation. If it is a reversed polarity trace
associated with a specific receiver location, then two
consecutive crosscorrelations using contiguous pairs of
receiver locations will contain a dominant trough.
The crosscorrelation technique described below was
used for the computation of residual static corrections
prior to the introduction of more automated programs
in the late 1960s and early 1970s (e.g., Disher and
Naquin, 1970; Ferree and Miller, 1972; Judson and
Sherwood, 1973; Martin, 1978; Silverman, 1978; Stevens,
1982). The technique is also described in Section 7.3
where it is used in an automated or semi-automated
mode for the evaluation of residual static corrections.
Raypaths from a source location S to receiver locations R1, R2, and R3 are shown in Figure 5-24 for a twolayer model with velocities V1 and V2. If the refraction
arrivals recorded at R1 and R2 are compared by a crosscorrelation technique, the measured time difference is
an estimate of the difference in traveltime between raypaths SABCR2 and SABR1. This has two key components: the time difference through the near-surface layer
at the two receiver locations and a refractor component,
the distance BC at a velocity of V2. There is also a noise
component. Section 5.6.2 shows that the two key components equate to the differential delay time between
the two receiver locations.
The effect of the noise component can be reduced by
averaging the time differences from a series of crosscorrelations involving different source locations but the

Fig. 5-25. Summation of individual crosscorrelations to


illustrate improvement in the signal-to-noise ratio. (Data
courtesy of Digicon Geophysical Ltd. )

same pair of receiver locations. For the averaging procedure to be valid, the refractor must not change as a result
of the different source-to-receiver offsets used. In addition, the receiver-to-receiver distance must be constant
for different source locations, which is not necessarily
the case for crooked-line or 3-D recording. As an alternative averaging procedure, the crosscorrelations can be
summed and then picked (Figure 5-25). This has the
potential advantage of improving the signal-to-noise
ratio prior to the picking process, or at least reducing the
contribution of the poor quality crosscorrelations. Other
variations include crosscorrelations of an individual
trace against a model trace, such as a common-receiver
stack (e.g., Musser et al., 1991), which can be used for the
analysis of refraction and reflection data (described in
Section 7.4.3.4).
If one of the two averaging procedures is repeated for
all pairs of consecutive receiver locations along the line
(e.g., from R2 to R3, R3 to R4, R4 to R5, and so on), a relative refraction arrival-time profile for the line can be
generated. This is achieved by assigning an arbitrary

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


S4

S1

R1

S2

R2

S3

Fig. 5-26. Plan view of short line segment with source


locations offset from the main receiver line.

value such as 0 to the first receiver location and then


integrating successive differential times estimated from
the crosscorrelations. The resulting profile is equivalent
to the continuous relative delay time profile described in
the Gardner method (Section 5.6.6).
If a reversed polarity receiver location is present, two
consecutive crosscorrelations contain a dominant
trough. If the trough is not picked, which is often the
case, one of the peaks on either side of the trough is
picked. A cycle skip in the relative refraction time profile
is thus likely if both the values picked are either positive
or negative times with respect to the time of the trough.
In practice, crosscorrelations are normally performed
on reduced traveltime data, that is, after removal of the
refractor velocity component, or refractor moveout
(stepout). The refractor velocity term in the crosscorrelation then becomes a velocity error term. Errors associated with different receiver-to-receiver distances for
crooked-line or 3-D recording are normally minimized
to an acceptable level; this is because the correct sourceto-receiver offset is used for the refractor moveout.
However, the recording geometry may be such that
there is no common element to the refracted raypaths
that are to be compared by the crosscorrelation procedure. I review this by first examining situations in which
a common element exists. In Figure 5-24, there is a common element in the near-surface layer at the source (raypath SA) and in part of the refractor (raypath AB) for
receivers R1 and R2. In the plan view shown in Figure
5-26, which corresponds to the layout in Figure 5-24,
sources located at S1 and S2 are in-line with the two
receivers at R1 and R2. Thus, raypaths from the source to
the two receivers are common over much of their length.
This is not the case when the source is offset from the
line, as shown for source locations S3 and S4 in Figure
5-26. The travel paths through the near-surface layer at
the source and along the refractor are now different for

169

the two receiver locations. Similarly, the raypaths


through the near-surface layer up to the two receiver
locations at R1 and R2 are different for source locations
S3 and S4. For the in-line case, as shown with sources at
locations S1 and S2, the receiver raypaths are identical.
Thus, whenever the receiver line is not straight, or
when source locations are offset from the line, the crosscorrelation times between two consecutive receiver locations do not necessarily give the same differential times
for different source locations. Under these conditions,
averaging is no longer strictly appropriate. The individual time differences must be estimated and used in an
interpretation technique that accounts for the threedimensional nature of the near surface, as described in
Sections 5.6.10 and 5.6.11. In practice, however, there are
many situations when the differences in raypaths, and
hence differential arrival times, are small and the averaging procedure described can be used on all the data or
on a substantial subset of it.
If two directions of recording are used (split recording), the crosscorrelation procedure must be performed
separately for the two halves of the spread. This is
because the velocity term changes sign with the direction of recordingin one direction, correlations will be
from a long to a shorter offset trace, and in the other
direction, from a short to a longer offset trace. However,
this difference can be incorporated into the averaging
procedure and it can also be used to estimate the refractor velocity. In the Gardner method and other delay time
techniques (Section 5.6.6), the divergence between forward and reverse profiles is used to update the refractor
velocity from the initial estimate.
To overcome the problem associated with one poor
receiver location, which generates two consecutive poor
quality crosscorrelations, crosscorrelations can be generated between alternating locations, such as R1 and R3, R2
and R4, R3 and R5, and so on. These can be used to check
the validity or QC of the contiguous set of values, to skip
a poor receiver location, and to form separate odd and
even numbered receiver profiles. Gelius et al. (1984)
indicated that the gap, or number of receivers to be
skipped, might need to be increased further on problem
data sets. This technique may also be beneficial if
reversed traces are present. For example, if the crosscorrelations of R2 to R3 and R3 to R4 both indicate a dominant trough but R2 to R4 shows a dominant peak, this
indicates that the data at R3 are polarity reversed.
The technique described for the receivers can be
modified to obtain an equivalent relative refraction
arrival-time profile for the source locations. This
requires reversing the role of the source and receiver in
Figure 5-24. That is, traces from two source locations
recorded by the same receiver are crosscorrelated. The
data can also be crosscorrelated on a common-offset

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

170

Static Corrections for Seismic Reflection Surveys

Fig. 5-27. Quality control of refraction arrivals with spikes superimposed on the data at the picked time: (a) wiggle trace
display (Ketelsen and Fromm, 1982; preprint); (b) VA/WT Display.

basis (e.g., Coppens, 1985; Martin and Fenley, 1985). This


has the advantage of removing the effect of any velocity
error term, but the differential time between the two
traces must then be split into its source and receiver
components. For multifold recording, many commonoffset comparisons can be performed. The average difference between two source locations with different
pairs of receiver locations is close to the required sourceto-source difference because the receiver terms average
out to nearly zero (Martin and Fenley, 1985). For
crooked-line or 3-D recording, the velocity error term
can be minimized by ensuring that the common offsets
are within a small value of each other; this value should
be selected so that the refractor traveltime difference is
no more than a few milliseconds after the application of
refractor moveout.
Holst et al. (1985) proposed an interactive approach
based on an optical correlation of data sets. Two refractor moveout corrected source records aligned on a common-receiver-location basis are displayed, and the second source record is time shifted to give the best match
of the refraction arrivals across the recording spread.
The necessary time shift is the differential refraction
arrival time between the two source locations. Similarly,
receiver gathers are displayed on a common-sourcelocation basis, and the time shift to align the refraction
arrivals is the receiver-to-receiver differential refraction

arrival time.
As a QC procedure, various simple comparisons can
be performed on the differential refraction arrival times
and on the relative refraction arrival-time profile. After
accounting for refractor moveout, the time shift between
two traces can be checked to ensure that it is within a
specific threshold. Time shifts from crosscorrelations of
noncontiguous locations should be checked to ensure
they are close to the sum of the relevant time shifts from
crosscorrelations of consecutive locations. Major differences point to a timing error in one or more of the crosscorrelations. For split recording, the relative refraction
arrival-time profile from the two directions of recording
should show a broad similarity because the only major
difference is the velocity error, which is opposite in sign
for the two directions of recording. Source and receiver
profiles should also be similar, although their high-frequency (short-wavelength) component is likely to be
different. Comparisons should also be made between
the profile and any available near-surface information,
such as an elevation profile.

5.5.6

Display of Refraction Picks

Refraction data should be displayed after the refraction arrivals have been picked as a QC procedure, either
in the form of seismic data or picked refraction arrival

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


1

REFLECTION
DATA
FROM
RECEIVER
6

REFRACTION
DATA

171

0.0

0.0

0.5

0.5

OFFSET
IN
FEET
690

1.0

1.0

RECEIVER
5

2000

3640

5280

6920

8400
CDP POSITIONS FOR SOURCE
OVER LOW VELOCITY ZONE

RECEIVER OVER ANOMALY

Fig. 5-28. Offset panel with six offsets showing the response of low-velocity zone close to the water bottom; the
position of a source and receiver location close to the anomaly is shown (Fulton and Darr, 1984).

times. The display format should allow for easy identification of spurious or wild picks, anomalous source or
receiver locations, variations in refractor velocity, and
anomalies relating to incorrect or wrongly annotated
field geometry.
The seismic data can be viewed as individual records
with a spike added to the trace at the time of the pick, as
shown in Figure 5-27. This allows for a detailed analysis
of individual traces but does require that several records
are analyzed to see if a specific receiver location is anomalous. In some cases, the refractor moveout (stepout) is
removed from the data prior to the display so that the
refraction arrival is nearly constant in time. Similar displays can be produced that have the picked times applied
to the data. All refraction arrivals should then be aligned
at a constant time, and any anomalous values, such as
those due to a cycle skip, can be readily identified.
To identify anomalous source or receiver locations, it
is advantageous to see data from many source and
receiver locations along the line in a single display of a
convenient size. This can be achieved by showing a
short segment of the data, normally limited to a few
hundred milliseconds, with the gate or window starting
just before the refraction arrival. Fulton and Darr (1984,
1985) proposed the offset panel as a way of displaying
data to allow the interpreter to easily review the data.
Data from a range of offsets, almost equally spaced
across the recording spread, are plotted one beneath
another in their CMP position, as done in Figure 5-28. If
the recording geometry is regular, diagonal alignments

correspond to features associated with a specific source


or receiver location.
A shallow anomaly affects the traveltimes and amplitudes of all offset displays, at both source and receiver
locations, whereas a deeper anomaly affects only the
longer offset data (the diagonal alignments are also
shown in Figure 6-27 which shows reflection patterns
from a surface feature). The type of display in Figure
5-28 and the constant-offset refraction arrival displays in
Figure 5-29 can be used to aid in the selection of an
appropriate window prior to automatic picking of
refraction data, or as a QC procedure on the picked
arrivals. In areas of poor signal-to-noise ratio, a limited
amount of stacking prior to data display may be beneficial, although this is not the case if large trace-to-trace
static variations are present, as both noise and signal are
then attenuated by the process.
Displays can also be produced on a common-source
or common-receiver basis, normally after the application of refractor moveout. Figure 5-30 shows data after
removal of the refractor moveout term, displayed as
source records plotted against receiver location, where
the source is located in the center of a 96-trace recording
spread. In this example, each trace segment is less than
300 ms long so that a large amount of data can be formatted into one display. Only a portion of the line is displayed because the data continue below the base of the
figure at source location 272. In this figure, 38 source
locations are plotted before the display wraps around,
with the 39th location at the top of the display. The fig-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

172

Static Corrections for Seismic Reflection Surveys


Receiver Station Position
500

400

300

200

100

Time (s)

0.1
0.3
0.1
0.3
0.1
0.3
Fig. 5-29. Discrete offset refraction arrival displays. (Data courtesy of Western Geophysical.)

ure shows a change in source frequency between source


locations 275 and 272. Slower velocities are observed on
the inside traces at several source locations, such as 197
to 188; this may indicate the velocity of either a slower
refractor or the direct arrival.
This type of display allows the interpreter to see time
variations that are associated with a receiver for all
source locations (such as the anomaly between receivers
258 and 259 in Figure 5-30), and it allows easy identification of anomalously noisy receivers. In the complementary display, receiver gathers are plotted against
source locations (e.g., Farrell and Euwema, 1984a).
Similar source and receiver displays can be produced
in which the picked times are applied to the data such
that all refraction arrivals are aligned at a constant time
and any mispicks and cycle skips are easy to identify. In
addition, displays can be generated after the application
of datum static corrections as an approximate QC check
of these corrections. The approximation arises because
datum static corrections for reflection data do not necessarily align refraction arrivals. This is because the nearsurface raypaths are not vertical, which is the assumption for reflection datum static corrections; these differences are also discussed in Sections 5.5.7.5 and 7.12.
The common source or receiver data can also be
summed after refractor moveout to form common
source and receiver stacks. These will be poor quality if
large trace-to-trace time shifts are present and are thus
used mainly after the application of picked times or
datum static corrections to the data. These stacked displays allow for quick evaluation of data quality along
the line and give an indication of the accuracy of the
picking process. However, they cannot be used to critique the refractor velocity or to detect a change of
refractor within the offset range analyzed.
The picked refraction arrival times can also be displayed in a variety of formats. The conventional
timedistance display, in which the picked times are
plotted against offset or receiver location, is typically
used for data from a weathering or LVL survey (see

Figure 5-13). Similar displays of the refraction arrival


times can be produced from a seismic reflection crews
data. In the example in Figure 5-31, most of the traveltime curves parallel one another, although there are
many mispicks in the unedited version (Figure 5-31a).
After editing, almost all of these have been identified and
removed, as shown by the updated display in Figure
5-31b. This shows several receiver anomalies, such as
those between receiver locations 111 and 112 and at 119.
In split recording, two directions of recording are
used and two such displays are normally produced so
that they do not have a series of crossing events. If only
a few isolated shots are in the reverse direction, however, they are often included as part of the main display for
the line.
A similar display can be generated after removal of
the refractor moveout term, as shown by Chun and
Jacewitz (1980, 1981, 1983), which has a format similar to
that of the time-gated seismic data in Figure 5-30. A profile of picked times for a specific offset was shown in
Coppens (1985).
Figure 5-32 illustrates several ways of displaying
refraction arrival times and how source, receiver, and
velocity anomalies manifest themselves in source-,
receiver-, and offset-ordered displays. In Figure 5-32a,
the data are displayed as source-ordered arrival times
plotted against receiver locations. Receiver anomalies
are observed at locations 14 and 15 and a time anomaly
between receiver locations 17 and 22. The offsets to the
right of this display all show anomalous times, which is
probably associated with a velocity error. This is clearer
in Figure 5-32c, in which the data are displayed as
source-ordered arrival times plotted against trace offset.
Offsets 17 to 24 show a change in slope and hence velocity, indicating the presence of a faster refractor. The
receiver anomalies aligned vertically in Figure 5-32a are
observed as diagonal alignments in Figure 5-32c. In the
receiver-ordered display of Figure 5-32b, the arrival
times are plotted against source position, and source
anomalies are shown at locations 15, 19, and 21. Thus,

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

Fig. 5-30. Refraction arrivals after removal of refractor moveout or stepout; source-ordered display plotted against
receiver location (annotated values in display are source location numbers).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

173

174

Static Corrections for Seismic Reflection Surveys

Fig. 5-31. Timedistance display of refraction arrival times plotted against receiver position: (a) raw picks; (b) edited
picks (Ketelsen and Fromm, 1982; preprint).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

175

(b)

(a)
1

10

Receiver location
30
20

50

40

11

Source location
21
31

41

1
1
5

10 ms
11

Source location

10
21
15
31

Receiver location

20
41

51

25

10 ms

30

(c)
1

Offset (group intervals)


6
18
12

24
35

40

Source location

11

45

21

50

31

41

10 ms
51

Fig. 5-32. Quality control display of refraction arrival times after removal of refraction moveout or stepout:
(a) source-ordered times plotted against receiver location; (b) receiver-ordered times plotted against source location;
(c) source-ordered times plotted against trace offset.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

51

176

Static Corrections for Seismic Reflection Surveys

(c)

0
100ms
200ms
300ms
400ms

(d)

0
100ms
200ms
300ms
400ms

Fig. 5-33. Display of refraction arrival times: (a) near-surface model showing position of a source and receiver anomaly
and refracted raypaths; (b) first-arrival-time surface; (c) receiver projection of the arrival times; (d) source projection of
the arrival times. Variables: V0 is near-surface velocity and V1 is refractor velocity (Chun and Jacewitz, 1981; preprint).

each of the three displays focuses on the identification of


specific anomalies: a source-ordered display versus
receiver location for receiver anomalies, receiverordered versus source for source anomalies, and sourceordered versus offset for velocity anomalies.
An isometric display of picked times was shown by
Farrell and Euwema (1984a), and an iso-offset time display by Morack et al. (1984). A first-arrival-time surface
was shown by Chun and Jacewitz (1981, 1983), which is
illustrated in Figure 5-33b for the model data in Figure
5-33a. Chun and Jacewitz also proposed a display in
which the time picks are projected onto the source and
receiver axes to produce a scatter plot (Figures 5-33c and
d). Here, the picks at one source location show all time
values for its receiver locations. This can then be com-

pared to a similar scatter plot after application of static


corrections, which should show much less scattering of
the values.
I mentioned in Section 5.4.3 that refraction arrivals
were often used, especially in 3-D surveys, as a means of
verifying that the geometry used for each source was the
same as that recorded in various logs by the acquisition
crew. An approximate near-surface model is generated
and first-arrival times are then modeled using the
assumed recording geometry and the source and receiver coordinates. These times are then compared with the
data, often facilitated by superimposing the modeled
picks onto the recorded data, in a display similar to one
of those in Figure 5-27. Areas where the match is poor
are then investigated to ascertain whether the difference

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


is caused by a poor near-surface model or is due to an
error in the reported or logged recording arrangement
or survey coordinates (e.g., Luzietti et al., 1995).

5.5.7

Refraction Arrival-Time
Adjustments

Various corrections may need to be applied to the


refraction arrival times prior to the interpretation stage.
First-break times are normally converted to surface-tosurface times, or near-surface corrections are computed
and applied to adjust the times to a reference datum.
When peaks, troughs, or zero crossings are picked, or
when a relative approach is used such as the crosscorrelation technique (Section 5.5.5), these can be converted to
equivalent first-break times. This requires an estimate of
the time from the phase picked to the first-break time,
taking into account frequency variations along the line, a
procedure described in Section 5.5.1. Alternatively, a relative profile can be used in the interpretation and converted to absolute time and depth using external control,
such as information from an uphole survey or from a few
refraction profiles that have been converted to first-break
times. Various factors that may need to be applied to
first-break times can be summarized as follows:
1. A time break correction is applied when the time
of detonation, or initiation of the seismic wave,
does not correspond to time zero (see Section 3.9).
2. Any delay associated with the recording instruments should be measured and removed.
3. Uphole corrections are required for dynamite surveys when the interpretation technique assumes
that the refraction arrival times are surface referenced, that is, surface-to-surface times are
required. Corrections for the source in or below
the weathered layer are covered in Section 5.5.7.1.
4. If source or receiver arrays are used in the field,
or if the source is offset an appreciable distance
from the line, a correction factor may need to be
applied (see Sections 5.5.7.2 and 5.5.7.3).
5. If the line is crooked or is part of a 3-D survey,
various offset corrections must be applied (see
Section 5.5.7.4).
6. Weathering and elevation corrections are required
when the interpretation technique assumes that
the refraction arrival times are referenced to a
datum (see Section 5.5.7.5).
Many of these corrections require the refractor velocity; however, because the corrections are required prior
to the full refraction interpretation, only an estimated
value is usually available. In most cases, this should be
sufficiently accurate for the correction concerned. If nec-

177

essary, a sensitivity check can be made or the corrections


recomputed after establishing the correct refractor
velocity and noting the magnitude of the differences.
Only in extreme cases is it necessary to redo the refraction interpretation with an updated set of corrected
arrival times.

5.5.7.1 Surface-to-Surface Times


When refraction information is used to derive a nearsurface model, it is normal practice to convert the firstbreak times to surface-to-surface times. This procedure
is applicable when the source is below the surface, such
as in a deep-hole dynamite survey. In some interpretation techniques, it may not be necessary to apply this
correction to all the data, just on those profiles used to
establish depth control along the line. These corrections
should also be carried out for shots used to check reciprocal time ties. Different types of corrections are
required depending on whether the shot is in or below
the weathered layer; both of these cases are detailed
below. An additional correction is required for the direct
arrival.
In the Blondeau method (Section 5.6.7), the near-surface layers are assumed to have an instantaneous velocity proportional to a power of the depth, and it is customary to apply a simple uphole correction. This correction assumes vertical travel through the near-surface
layers, and it is followed by a subsequent modification
to account for the actual raypath.
Figure 5-34 shows a near-surface model with a source
at S in the weathered layer at depth ds below surface
location S. The travel path for a refracted arrival from
source S to receiver R is SBCDR. If this source is now
placed at the surface at S, the new travel path is SCDR.
To simulate a surface source, the required time correction (tc) added to first-break times recorded at receivers
such as R is thus given by
tc =

SC S B BC

,
V1 V1 V2

(5.36)

where V1 and V2 are the velocities in the weathered and


subweathered layers, respectively.
If the interface between the two layers is fairly flat
(less than about 10), SB and BC are almost equal to AC
and SA, respectively. Under these conditions, equation
(5.36) can be simplified to
tc =

SA S A

.
V1
V2

Using simple geometry, we can express this as

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

178

Static Corrections for Seismic Reflection Surveys


S

Xi
S

ds
S

dw

Di

ds

V1

V1
C

c
V2
B

V2

Fig. 5-34. Source-to-receiver raypath to illustrate the


uphole correction for refraction data with the source in
the weathered layer. Variables: Xi is source-to-receiver
offset, ds is source depth, Di is distance from S to R, c
is critical angle, V1 is weathered layer velocity, and V2 is
subweathered layer velocity.

Fig. 5-35. Source-to-receiver raypath to illustrate the


uphole correction for refraction data with the source
below the weathered layer. Variables: dw is thickness of
weathered layer; others as in Figure 5-34.

ti = ti
tc =

ds
1 ds tan c
,

cos c V1
V2

(5.37)

where c is the critical angle.


Equation (5.37) can be simplified using Snells Law,
defined by equations (5.1) and (5.2):
ds cos c
,
V1

(5.38)

tc = tuh cos c ,

(5.39)

tc =
or

where tuh is the uphole time for location S, that is, the
time from S to S. In areas characterized by a large velocity contrast between V1 and V2, the cos c term in equations (5.38) and (5.39) can often be neglected. For example, with velocity contrasts in excess of 3.0, cos c is
greater than 0.95.
The direct arrival times also require modification if
the source is located in a deep hole. When the receiver is
located at R, as in Figure 5-34, the actual distance traveled (Di) by the direct arrival in the weathered layer is

Di = X i2 + ds2

1/ 2

where ds is the source depth and Xi is the horizontal distance from source location S to receiver R. The direct
arrival time (ti) for an equivalent source located at the
surface is thus given by

(X

Xi
2
i

+ ds2

1/ 2

(5.40)

where ti is the observed time from the source in the


weathered layer at S. Equation (5.40) assumes that the
velocity does not change spatially or with depth; if the
velocity does change, equation (5.40) represents an
approximation of the correct adjustment. In most areas,
however, it is not necessary to compute a more accurate
correction, which would require ray tracing based on a
near-surface velocity field.
Figure 5-35 illustrates a near-surface model that
incorporates a source S below the weathered layer at
depth ds below surface location S. The travel path for a
refracted arrival from source S to receiver R is SDR. If
this source is now located at the surface at S, the new
travel path becomes SBCR. Points C and D represent the
same location on the refractor if the shot is located a
small distance below the base of the weathered layer.
This situation is more complex than when the source is
in the weathered layer, as just described. Another example of raypaths from a deep-shot hole was shown in
Figure 4-17, in which the dip of the interfaces was seen
to alter the actual raypath.
When the recording distance is much greater than the
depth of the hole below the weathered layer (that is, the
distance SR in Figure 5-35 is much greater than AS), the
distance from the shot at S to point D on the refractor is
almost equal to the distance AD. To simulate a surface
source at S, the required time correction (tc) to add to
first-break times recorded at receiver R is thus given by
a relationship similar to equation (5.36):
tc

SB AB

.
V1 V2

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

(5.41)

Chapter 5Refraction Surveys


Using the same procedure as in the derivation of equation (5.38), we can express equation (5.41) as
tc =

dw cos c
,
V1

(5.42)

where dw is the weathering thickness at location S and


c is the critical angle. Alternatively, equation (5.42) can
be rewritten as

d dw
tc = tuh s
cos c ,
V2

rection must thus be added to the first-break time to


obtain an adjusted time that corresponds to the trace offset. If a later phase of the first arrival is picked, such as a
peak, trough, or zero crossing, then an array correction
is generally not necessary, assuming that the picked
time corresponds to an average output from the trace.
The array correction (ta) to be added to first-break
arrival times from a refractor is given by
ta =

(5.43)

where tuh is the uphole time from the source at location


S, the time from S to S.
Equations (5.42) and (5.43) both require an estimate of
the weathering thickness dw, one of the parameters that
still has to be estimated from the seismic data. However,
it may be possible to obtain an estimate from the
drillers report, although its accuracy depends on the
drilling fluid, the nature of the change between the
weathered and subweathered layers, and the drillers
expertise. A more complete description of near-surface
corrections for deep-hole recording can be found in
Knox (1967).
When the hole depth is equal to the weathered layer
thickness, that is ds = dw, equations (5.42) and (5.43) are
identical to equations (5.38) and (5.39). With the source
below the weathered layer, a direct arrival correction is
not required because no direct arrivals through the
weathered layer are recorded.
If a delay time interpretation technique is used (see
Sections 5.6.2 and 5.6.6), delay times without uphole
time adjustments represent delay times at receiver locations where the source is located at or just below the
weathered layer or refractor. Thus, with a delay time
approach, it may not be necessary to compute the surface-to-surface times for all shots. However, it is still
required for those shots used to establish depth control
along the line, although depth control can be achieved
by an external source of information, such as uphole
surveys.

5.5.7.2 Array-Length Correction


If arrays are used for either the source or receiver, a
time correction may need to be applied to the refraction
arrival times. Array related amplitude and time distortion effects on the refraction arrival were described in
Section 5.4.3. If a first-break time is picked, it is normally assumed that it is recorded by the closest source element to receiver element; this is a shorter distance than
the offset for the group, which corresponds to the source
array center to receiver array center distance. A time cor-

179

SL + RL
,
2Vr

(5.44)

where Vr is the refractor velocity and SL and RL are the


apparent source and receiver array lengths, which refer
to lengths projected onto the line joining the source and
receiver locations. For example, RL = 0 if the receivers
are in a line perpendicular to the source-to-receiver
azimuth; the same receiver array will have a different
apparent array length for other source locations if the
source-to-receiver azimuth changes. Thus, the apparent
array length is less than or equal to the actual or physical array length. Variable corrections at a given location
are more likely to be a factor on crooked-line and 3-D
surveys than on a straight-line 2-D survey.
In many cases, especially in short arrays, the correction defined by equation (5.44) is small, but it is often
large enough that it should still be applied. For example,
with array lengths of 15 m for both source and receiver
and a refractor velocity of 2000 m/s, the correction is 7.5
ms. With variable source-to-receiver azimuths, the correction can change from 7.5 ms down to 0 ms for the
broadside case. When the intercept-time method
(Section 5.6.1) is used to interpret the data, an alternative
approach for straight-line recording is to plot the data at
the appropriate distance from the source to the receiver.

5.5.7.3 Source-Offset Correction


If the source is offset an appreciable distance from the
receiver line, a correction factor may need to be applied
to the refraction arrival times to compensate for the
increased distances traveled by the direct and refracted
arrivals. However, if delay times are being computed,
the actual source-to-receiver distance is used in the computation of the delay time (see Sections 5.6.2 and 5.6.6)
and no specific offset correction is required. This is also
the situation for the inversion, time term, and tomographic methods described in Section 5.6.9. However, if
composite timedistance curves are generated, or a simple zero-offset intercept time is computed (see Section
5.6.1), a correction factor should be applied to the refraction arrival times. The necessary adjustment (tc) to convert the observed times to those that would have been
obtained if the source were located on the receiver line is

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

180

Static Corrections for Seismic Reflection Surveys

tc

(H
=

+ X i2

1/ 2

Xi

Vr

(5.45)

where tc is the correction to be subtracted from the


refraction arrival times, H is the perpendicular offset of
the source from the receiver line, Xi is the trace distance
along the line, and Vr is the refractor velocity. However,
as noted in Section 5.5.5, when the source is offset from
the line, refraction raypaths are no longer common from
a source to different receiver locations within the recording spread. This means that the above correction and
any subsequent 2-D interpretation technique assumes
that no lateral changes in velocity occur, either of the
refractor or the near-surface layers. I discuss these points
further in Sections 5.6.10 and 5.6.11 on crooked-line and
3-D interpretation.
An alternative approach, especially when the interpretation technique requires the computation of the
zero-offset intercept time, is to plot the data at the appropriate distance from the source. That is, the data are plotted at the trace offset rather than at the in-line distance of
the receiver location; the trace offset (Di) is calculated by

Di = H 2 + X i2

1/ 2

(5.46)

For the direct arrival, an expression similar to equation (5.45) is used, but it requires a satisfactory estimate
of the direct arrival velocity. It is often more practical just
to plot the data at the appropriate offset on a timedistance display because this approach is required to estimate the direct arrival velocity prior to computing the
time correction.

5.5.7.4 Crooked-Line and 3-D Recording


For crooked-line and 3-D data, the actual trace offset
is used in delay time computations (see Section 5.6.2),
inversion, time term, tomographic methods (Section
5.6.9), and in some cases, plotting data prior to computation of the zero-offset intercept time. The trace distance (Di) is computed from the source and receiver survey coordinates by

Di = (Xs X r ) + (Ys Yr )
2

2 1/ 2

(5.47)

where Xs, Ys, Xr, and Yr are the x and y coordinates of


the source and receiver locations, respectively.
Specific interpretation techniques, such as those
described in Sections 5.6.10 and 5.6.11, are used for
crooked-line and 3-D surveys. These take into account
that when the source is offset from the line, refraction raypaths are no longer common from the source to different
receiver locations within the recording spread (Section

5.5.5). This implies that a 3-D interpretation of the near


surface is required. In addition, crooked-line data can be
interpreted as 2-D data, but this assumes a 2-D subsurface
with no lateral changes in refractor velocity nor any local
changes in near-surface layers adjacent to specific source
or receiver locations. Several refraction interpretation
techniques that can be used with these assumptions are
described in Section 5.6.10. The array-length correction,
which may be required to modify the first-break times, is
not constant for a specific source or receiver location in
crooked-line or most 3-D surveys because it depends on
the source-to-receiver azimuth and the layout of the array
(see Section 5.5.7.2).

5.5.7.5 Static Corrections for Refraction Data


The computation of datum static corrections for a
reflection survey from a near-surface model assumes
that the raypaths in the near-surface layers are vertical.
Section 6.2 discusses areas where this assumption is
invalid, such as complex overburden or highly irregular
deep water-bottom topography. Alternative techniques
of applying near-surface time corrections, such as waveequation datuming and time- and offset-variant shifts
(dynamic static corrections) derived from ray tracing,
are described in Sections 6.2.3 and 6.2.4.
In refraction surveys, however, raypaths are rarely
close to vertical in near-surface layers. This is because
the incident angle to the target refractor (the critical
angle) is likely to be large. The required static corrections
for refraction data take this into account. The static correction depends on the critical angle and therefore on
both the near-surface and refractor velocities. Because
the static corrections are refractor dependent, when several refractors are mapped, an equivalent number of static corrections are required. I discuss the difference
between reflection and refraction datum static corrections in the context of residual static corrections in
Section 7.12.
Section 5.5.7 stated that refraction first-break times
are generally converted to surface-to-surface times or
near-surface corrections are applied to adjust the times
to a reference datum, although this depends on the
interpretation approach used. The interpretation techniques that require the times to be converted to a reference datum are mainly for mapping refractors beneath
the top layer; however, there are occasions when this is
applicable to near-surface surveys.
Static corrections for refraction data can be subdivided into a weathering correction and an elevation correction;
these are the same divisions used for reflection data
datum static corrections described in Section 3.1. Using
the sign convention defined in Section 3.1, a negative
time correction reduces the refraction arrival times.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Figure 5-36 shows a source A at the base of a weathered or low-velocity layer with velocity V1 and thickness hR beneath a receiver at location R. The weathering
correction adjusts the traveltime measured at location R
to simulate the arrival time at point G, located at the
base of the weathered layer vertically below R. The
actual travel path for an upcoming ray from the lower
refractor with velocity V3 is HFR; the required travel
path starting from point H is HJG. The time correction
(twR) to be applied to the refraction arrival time at location R to remove the weathered layer is thus given by
HF FR HJ JG
+
tw R =
+
+
.
V1 V3 V2
V2

(5.48)

HF JG
and
HJ FG.
Using these approximations and a procedure similar to
that used to derive equation (5.38), we can express equation (5.48) as
hR cos 1
,
V1

ES

R
dS

ER

V1

hR
F
G

A
V2

Ed

B
c

H J

Reference
plane

V3

If we assume that the base of the weathered layer is


almost flat at point G, then

tw R =

181

(5.49)

Fig. 5-36. Near-surface model and source-to-receiver


raypath to illustrate static corrections for refraction data;
datum or reference plane below the base of the weathered layer. Variables: ES and ER are elevations of source
S and receiver R, hR is thickness of weathered layer at R,
Ed is elevation of reference plane, V3 is velocity of lower
refractor, c is critical angle, and 1 is incidence angle in
top layer for refracted ray from layer 2 to 3; others as in
Figure 5-34.

If the refractor is fairly flat, SC is nearly equal to BD,


and equation (5.50) can be simplified to

where
1 = sin1 V1/V3.

te S =

I show in Section 5.6.2 that equation (5.49) defines the


delay time of the weathered layer (or low-velocity layer)
for a refractor with velocity V3. There is usually an
equivalent correction at the source location, although
this is not the case for the example in Figure 5-36
because the source is located just below the base of the
weathered layer.
The elevation correction is designed to adjust the
arrival times to simulate refraction arrival times that
would have been seen if the source and receiver were
located on the reference or datum plane, which can be
located above or below the base of the weathered layer.
Thus, the elevation correction adjusts the traveltime for
the source at A and the pseudo-receiver at G on the base
of the weathered layer (Figure 5-36) to simulate the
recording at R from a source at S. At the source location, the original travel path for the downgoing ray to
the refractor is AD. With the source located at S on the
reference plane, the equivalent corrected travel path to
point D is SCD. Thus, the time correction (te ) to be
S
applied at any receiver from source S is
te S =

AD S C CD
+
+
.
V2 V2
V3

(5.50)

AB CD
+
.
V2
V3

Using the approach in the derivation of equation (5.38),


we can rewrite this as
te S =

hS cos c
,
V2

(5.51)

where hS is the thickness of the V2 layer between the


source location and the datum plane, the distance AS in
Figure 5-36.
A similar correction is required at the receiver location, where the upward-travel path JG is replaced by
JKR (Figure 5-36). Using the approach in the derivation
of equations (5.38) and (5.51), the receiver correction
(teR) is given by
te R =

hR cos c
,
V2

(5.52)

where hR is the thickness of the V2 layer between the


base of the weathered layer and the datum plane, the
distance GR in Figure 5-36.
As in the weathering correction derived earlier, the
elevation corrections defined by equations (5.51) and
(5.52) also represent delay times (Section 5.6.2). Using

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

182

Static Corrections for Seismic Reflection Surveys

ES

S
R
V1
S

Reference
plane

ER

corrected travel path is SC. Thus, the time correction


(teS) to be applied to refraction arrivals from a source at
S is given by

Ed

te S =

AD CD S C
.

+
V2
V3
V2

Using the same approximations in deriving equation


(5.51), equation (5.56) can be written as

V2

te S = +
KJ

DC
V3

Fig. 5-37. Near-surface model and source-to-receiver


raypath to illustrate static corrections for refraction data;
datum or reference plane above the base of the weathered layer. Variables as in Figures 5-34 and 5-36.

the elevations, source depth, and thicknesses shown in


Figure 5-36, we can express the elevation corrections
given by equations (5.51) and (5.52) as
te S =

(ES dS Ed ) cos c ,

(5.53)

V2

and
te R =

(5.56)

(ER hR Ed ) cos c
V2

(5.57)

where hS is the thickness of the V2 layer between the


source location and the datum plane, the distance SA in
Figure 5-37.
A similar correction applies at the receiver location;
the upward raypath JG is replaced by the raypath KR.
Using the same approach as for equation (5.57), the
receiver correction (teR) is given by
te R = +

hR cos c
,
V2

(5.58)

where hR is the thickness of the V2 layer between the


base of the weathered layer and the datum plane, the
distance RG in Figure 5-37.
Equations (5.57) and (5.58) are identical to equations
(5.51) and (5.52) except for the sign. This is to be expected because the reference plane is above the base of the
weathered layer in one case and below it in the other.

(5.54)

The total correction (tSR) for receiver R from source A


is the sum of the relevant weathering and elevation corrections:
tSR = tw S + tw R + te S + te R .

hS cos c
,
V2

(5.55)

This total correction can be extended to the multilayer


case. Delay times of the individual layers, which vary
depending on the velocity of the target refractor, are
computed and summed together.
Figure 5-37 shows the situation in which the reference or datum plane is above the base of the weathered
layer. The near-surface model is the same as that in
Figure 5-36 (except for the elevation of the datum plane),
and the labels are the same in both figures. The weathering correction is identical in both, that is, it is independent of the datum plane elevation.
The elevation correction adjusts the traveltime for
the source at A and the pseudo-receiver at G on the base
of the weathered layer, to simulate the recording at R
from a source at S. At the source location, the original
travel path for the downgoing ray to the refractor is AD;
with the source located at S on the datum plane, the

5.6

REFRACTION INTERPRETATION
TECHNIQUES

For the simple case where the refractor is planar and


the velocities in the overlying strata are spatially invariant, the corrected refraction times can be converted to
depth using one of the approaches outlined in Section
5.3. In practice, the geology is generally more complex
than implied by this simple case, but a reasonable depth
model can often be obtained using this simplistic
approach. This is especially true for many LVL surveys,
in which elevation and near-surface geology changes
along the recording spread are minimal. Examples in
Section 5.6.1 show when this approach, the intercepttime method, is applicable and when the near surface is
too complex for it to work.
Timedistance curves obtained from a seismic refraction survey contain information about the velocity and
structure of the refractor and, to a certain extent, the layers above it. It is important that the analysis of a specific refractor uses only arrivals associated with that refractor; this may not be the case if the incorrect offset range
is used because the arrivals may include those from a

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)

S1

S2

V1

V2

(b)

S1

S2

V1

V2

Fig. 5-38. Raypaths for a reversed refraction profile for


analysis by: (a) common-surface-location method;
(b) common-subsurface-location or common-refractoremergent-point method. Variables: V1 is near-surface or
weathered layer velocity and V2 is refractor velocity.

shallower or deeper layer. Section 5.3.6 showed that


refraction information cannot be converted to a unique
depth model because a hidden layer, low-velocity layer,
or velocity inversion may be present. Additional control
is needed, such as that obtained from a deep uphole survey, to restrict the number of possible solutions.
Over the years, many different methods of refraction
interpretation have been proposed; some of these are
more applicable to shallow refraction surveys, others to
deep targets. I noted in Section 5.1 that the differences
between some of the methods was fairly small. However,
it seemed appropriate to mention the highlights of most
methods that are currently in use or that have been used
in the past but on data that are still current (even though
they may no longer be used).
Refraction interpretation methods can be broadly
divided into two basic approaches: those in which the
data are analyzed at a common surface location and
those in which a common subsurface location, or common emergent point from the refractor, is used. The corresponding raypath diagrams for these two approaches
are shown in Figure 5-38. The common surface location
is point R in Figure 5-38a, where data are analyzed from
source locations S1 and S2. The common emergent point
is labeled E in Figure 5-38b, so that data are analyzed at
surface locations R1 and R2 from sources S1 and S2.
The concept of delay time is used in several methods
(see Section 5.6.2). Section 5.5 states that when refraction

183

data are stacked on a common-surface-location basis,


the traveltimes picked are equivalent to a delay time
profile. These traveltimes include the final stack and the
time shifts applied in the generation of the stacks.
The techniques commonly used for near-surface control include (or are based on) the ABC method (Section
5.6.3), Hagedoorn or plus-minus method (Section 5.6.4),
generalized reciprocal method or GRM (Section 5.6.5),
Gardner method (Section 5.6.6), Blondeau method
(Section 5.6.7), and the inversion, time-term (decomposition), and tomographic approaches described in
Section 5.6.9. The general wavefront technique
described in Section 5.6.8 has historically been considered a graphical method, and as such gives the interpreter a good understanding of the refraction method
and some of its limitations. It has recently been implemented as a data processing procedure by using a
downward-continuation technique on the recorded
wave field. Specific points about the interpretation of
data from crooked-line surveys, 3-D surveys, and
marine surveys are described in Sections 5.6.105.6.12.
Most of the methods are illustrated with a worked
example, and in some cases, the same data set is used for
several techniques. I briefly discuss the applicability of
the various interpretation methods to different types of
refraction surveys and near-surface conditions in
Section 5.6.13. A glossary of refraction interpretation
methods, which uses definitions from Sheriff (1991), is
included in the Chapter 5 Appendix.
In nearly all refraction interpretation techniques, an
assumption is made that the refracted ray travels in a
vertical plane, defined at the surface by the source and
receiver locations. This is not the case if there is significant cross dip on the refractor since an emergent point
from the refractor will be outside this plane. I discuss the
noncommonality of these raypaths in Section 5.6.10.
As an alternative to the two basic approaches summarized earlier (based on raypaths shown in Figure
5-38), the computed refraction times and subsequent
refractor depths can be assigned to the common-midpoint (CMP), midway between the source and receiver
(e.g., MacPhail, 1967; Rhl, 1995). The usage of this technique for refraction interpretation has been comparatively small. However, refraction data sorted in the CMP
domain are used to analyze residual static corrections
(see Section 7.12).
Most of the refraction interpretation techniques generate time information that is subsequently converted to
depth. Section 5.7 covers the required velocity control
for near-surface and refractor velocities, including comments on the refractor depth and near-surface velocity
ambiguity, as well as errors in the final depth profile and
how these can affect datum static corrections computed
from these profiles. As with any geophysical interpreta-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

184

Static Corrections for Seismic Reflection Surveys

(a)
SP C

SP A
80

Time (ms)

1640 m/s

1690 m/s

40
480 m/s

470 m/s

0
0

40

20

60
Distance along spread (m)

80

115

100

(b)
SP D
100

SP F
3070 m/s

2975 m/s

Time (ms)

80
60
420 m/s

40

400 m/s
1760 m/s

1780 m/s

20
0
50

100
Distance along spread (m)

150

205

(c)
SP K
120

SP L

Time (ms)

12000 m/s

SP M

7000 m/s

80

40

0
0

100

200

300 (0)
100
Distance along spread (m)

200

300

Fig. 5-39. Timedistance displays of reversed refraction profiles from near-surface surveys: (a) 115-m recording spread;
(b) 205-m recording spread; (c) data from two contiguous 300-m recording spreads.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Table 5-5. Velocities and Intercept Times for Data
in Figure 5-39a.

Near-surface velocity (m/s)


Refractor velocity (m/s)
Intercept time (ms)

Shotpoint A

Shotpoint C

470
1690
10

480
1640
8

tion, the refractor depth profile and associated velocities


can be compared with the original data, the refraction
arrival times, by forward modeling from the surface to
the refractor. Differences can be used to update the profiles, as shown in Sections 5.6.9.1 and 5.6.9.3 for inversion and tomographic methods.

5.6.1

Intercept-Time Method

The intercept time was defined in Section 5.3 as the


refraction arrival time at zero offset, where it was estimated by extrapolation of the timedistance curve with
the apparent velocity. This time can be used to compute
the refractor depth. The intercept-time method (also called
the intercept method) is applicable to areas where the
near-surface geology is comparatively simple and
where changes in static corrections along the recording
spread (trace-to-trace time shifts) are minimal. The technique is normally applied to data from a reversed profile. Cunningham (1974) showed how the method can
be adapted for single-ended refraction profiles without
having to reverse the role of sources and receivers to
generate a pseudo-reverse profile. This is not always
appropriate, especially if the locations are not collocated. Johnson (1976) discussed the interpretation of split
spread refraction data.
Figure 5-39a shows an example of picked times from
a reversed refraction profile which I interpret as a twolayer case. The velocities (Table 5-5) can be estimated
graphically or computed, for example, using a leastsquares fit criterion. Individual picked times normally
have random time errors due to noise and can also contain a localized bias or anomaly due to variations in the
near surface. The latter effect should be observed from
both directions of recording at a specific surface location
and must be taken into account when the appropriate
velocity line is drawn on the timedistance plot. An
average near-surface velocity of 475 m/s and refractor
velocity of 1665 m/s can be used in equation (5.7)
(Section 5.3.1) to convert the intercept times in Table 5-5
to refractor depths of 2.3 and 2.0 m at shotpoints A and
C, respectively.
The data example shown in Figure 5-39b indicates a
three-layer case. The velocities and intercept times are
given in Table 5-6. An average near-surface velocity of
410 m/s and a refractor velocity of 1770 m/s are used in

185

Table 5-6. Velocities and Intercept Times for Data


in Figure 5-39b.

Near-surface velocity (m/s)


Shallow refractor velocity (m/s)
Intercept time (ms)
Deep refractor velocity (m/s)
Intercept time (ms)

Shotpoint D

Shotpoint F

420
1760
14
2975
26

400
1780
15
3070
28

equation (5.7); the depths to the shallow refractor under


shotpoints D and F are then computed as 2.9 and 3.1 m,
respectively. The thickness of the second layer can be
estimated from these values using equation (5.14); the
depth to the second refractor is computed as 16.0 m at
shotpoint D and 17.3 m at shotpoint F. The resulting
near-surface model is shown in Figure 5-40 together
with the raypaths that correspond to the first arrivals
plotted in Figure 5-39b.
The raypaths in Figure 5-40 illustrate that refraction
arrivals are recorded only from two short segments of
the shallow refractor at either end of the recording
spread and that there is no reversed coverage. In other
words, first arrivals from the shallow refractor are not
recorded from both directions from a common segment
of the refractor. Thus, the estimated velocities are independent of each other and could both represent updip
or downdip velocities. In the latter case, the refractor
would be a syncline rather than the inclined plane estimated in the interpretation described earlier. This lack of
coverage is a common occurrence in LVL surveys when
more than one shallow refractor is present. This effect
can be minimized by taking the necessary steps to
ensure that all refractors are adequately mapped with
reversed coverage data; this can often be achieved with
additional shots.
In the case of the deeper refractor, the emergent raypaths in Figure 5-40 indicate that there is much better
coverage than on the shallow refractor. Reversed coverage data are recorded on about half the recording
spread, and one-way coverage extends almost to each
end of the spread. It is thus important to recognize the
assumptions inherent in the interpretation and the limitations in the resulting near-surface profile or model.
Two contiguous reversed profiles, each 300 m long,
are shown in Figure 5-39c. The reversed spread between
shotpoints K and L appears at first sight to be a threelayer case. However, the third-layer velocities are 7000
and 12,000 m/s, which are too high (the maximum
velocity seen at the earths surface rarely exceeds 7500
m/s). It is likely that both these velocities represent
updip velocities or are influenced by significant changes
in the near surface. It is thus possible that both the
apparent second- and third-layer velocities are associated with one layer.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

186

Static Corrections for Seismic Reflection Surveys

50

Distance along spread (m)


100

205

150

SP D
0

SP F

Depth (m)

One-way coverage

One-way coverage

10

Reversed coverage

15

Fig. 5-40. Refractor depth profile of data shown in Figure 5-39b interpreted using the intercept-time method (vertical
exaggeration x5); raypaths shown correspond with the refraction arrival times plotted in Figure 5-39b.

150

1650 m/s

1780 m/s

Time (ms)

100

53 ms

50

58 ms
1000 m/s

800 m/s

34 ms
17 ms

375 m/s

420 m/s

0
S-50

S-1

Recording spread
0

50

S1

S50

100 m

Fig. 5-41. Timedistance display from the LVL survey shown in Figure 5-13 annotated with apparent velocities and
intercept times.

The timedistance curves between shotpoints L and


M in Figure 5-39c show many anomalous points, probably indicative of significant near-surface changes in
the recording spread. Crooked-line recording could
also be a contributing factor, although this is not the

case here. I consider that both these profiles are too


complex for the simple intercept-time method and that
a different technique is required to define the near-surface model. I give a full interpretation of this data set in
Section 5.6.4.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


The LVL survey data shown in Figure 5-13 is redisplayed in Figure 5-41 and annotated with apparent
velocities and intercept times. In this figure, data from
the far-offset shots are used in the definition of the
apparent velocities from the deeper refractor recorded
by the short-offset shots. Various anomalous values are
seen in the figure, such as the time low at a distance of
about 70 m, especially from the deeper refractor. On this
data set, there is again no reversed coverage on the shallow refractor, which has a velocity of about 900 m/s.
The coverage on the deeper refractor is also poor, with
just the source locations at each end of the spread (the
short-offset data). However, it is considerably improved
with the additional off-end shots. Three layers are indicated, with velocities of about 400, 900, and 1700 m/s.
In interpreting a data set such as this, the objectives of
the survey must be considered. This may mean that a
full interpretation is required, with each anomaly investigated to produce detailed depth and velocity profiles.
Alternatively, the results might be used as control points
along the line when depth and velocity information is
required only at one location, normally the center of the
recording spread. In this case, one approach is to average the apparent velocities and intercept times from the
two recording directions. This is appropriate when the
refractor dips are small, such as in Figure 5-41. The dip
appears to be significant on the shallow refractor, with
intercept times of 17 and 34 ms at the two ends of the
profile. However, because the near-surface velocities are
low, this translates to a dip of less than 2.0. This simple
approach gives depths of 5.6 and 20.6 m for the two
refractors using equations (5.7) and (5.14). If a detailed
interpretation is required, a different technique should
be used because the intercept-time method is applicable
only to relatively simple near-surface configurations.
I discuss an alternative approach in Section 5.6.13 in
which the interpretation is performed in two stages. The
short-wavelength (high-spatial-frequency) variations
are analyzed first to remove the trace-to-trace variations.
This is followed by an evaluation of the long-wavelength component. This technique allow the intercepttime method to be applicable to more areas.

5.6.2

Delay Time Concept

I showed in Section 5.3 that the extrapolation of the


timedistance curve to zero offset with the apparent
refractor velocity gave the intercept time. This can be
used to compute the sum of the refractor depths below
the source and receiver. If the refractor is planar, simple
geometric relationships can be used to separate this sum
into its two components, the source and receiver depths.
The intercept time (t0) is expressed as

187

zR
V1

zS

V2

Fig. 5-42. Source-to-receiver raypath for a two-layer


model to illustrate the delay time concept. Variables: X is
source-to-receiver offset, zS and zR are depths of nearsurface layer at S and R; others as in Figure 5-38.

t0 = t i

Xi
,
Va

(5.59)

where ti is the arrival time at offset Xi and Va is the


apparent refractor velocity, estimated from a timedistance display of the refraction data. I showed in Section
5.3.3 that this was dependent on the true refractor velocity and the dip of the refractor. If the true refractor velocity is substituted for Va in equation (5.59), the resulting
intercept time is the difference between the actual traveltime and the time to travel the horizontal distance
from the source to the receiver at the refractor velocity.
The portion of this intercept time associated with one
end of the raypath was defined by Gardner (1939a, b) as
the delay time. Thus, the intercept time is the sum of the
source delay time and the receiver delay time. This concept
assumes that the refractor is flat where the refracted
wave enters and leaves the refractor. In practice, it is limited to situations in which the refractor dip is small, no
more than about 10.
The total traveltime (tx) from a source at S to a receiver at R, shown in Figure 5-42, is given by
tx =

SB BC CR
+
+
,
V1 V2
V1

(5.60)

where V1 and V2 are the near-surface layer and refractor


velocities. Equation (5.60) can be rearranged and summarized for gentle dips as follows:
SB AB AB BC CR CD CD
tx =

+
+

+
+
.
V2 V2 V1
V2
V2
V1 V2
Thus,
tx =

X
+ td S + td R ,
V2

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

(5.61)

188

Static Corrections for Seismic Reflection Surveys

60

60

V1

Elevation (m)

Elevation (m)

40

V1 (< V1)

20

S2

S1

S2

S1

40
V1

20
V2

V2 (< V2)

V2
0

20
S1

S2

40
S2
60

80

Fig. 5-43. Near-surface model with surface features and


associated reduced refraction arrival times; source locations at S1 and S2. Variables: V1 and V1 are near-surface
velocities and V2 is refractor velocity.

where tdS and tdR are the delay times at source location S
and receiver location R, respectively, and X is the distance from S to R, which is almost equal to the distance
AD.
The delay times for the source at S (tdS) and the
receiver at R (tdR) can thus be expressed as
td S

SB AB
=

,
V1 V2

Reduced traveltime (ms)

Reduced traveltime (ms)

20

40

60

80

S1

100

Fig. 5-44. Near-surface model with subsurface features


and associated reduced refraction arrival times; source
locations at S1 and S2. Variables: V1 is near-surface
velocity and V2 and V2 are refractor velocities.

(5.2) in Section 5.2.1, can be used to simplify and


rearrange this expression to give
V1td S

zS =

cos c

(5.62)

td R

(5.63)

Using simple geometry and assuming that the refractor


is flat near point B in Figure 5-42, we can express equation (5.62) as
td S =

(5.64)

or

and
CR CD
=

.
V1
V2

zS
1 zS tan c ,

cos c V1
V2

where c is the critical angle and zS is the refractor depth


at location S. Snells law, defined by equations (5.1) and

zS = td S

V22

V1V2
V12

1/ 2 ,

(5.65)

and similarly for the receiver depth zR from its delay


time tdR.
Thus, depths can be determined for the source and
receiver ends of the profile, providing the intercept time
can be separated into its component delay times. In the
special case when the interface is horizontal, each delay
time is half the intercept time.
In a reduced traveltime display, often used for refraction
data, an approximate refractor velocity is substituted for

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Va in equation (5.59). If the velocity used is that of the
refractor, a source-ordered display of the reduced traveltimes indicates changes in delay time at the receiver
locations. Figures 5-43 and 5-44 show the implications of
changes in near-surface and refractor velocities and
structure on a reduced traveltime display. To simplify
these displays, one of the reduced traveltime plots in
each figure is displaced in time.
Figure 5-43 shows the effect of surface features for
both forward and reverse directions of recording. The
true refractor velocity (V2) is used to compute the
reduced traveltimes. Because this velocity is much
greater than the near-surface velocity (V1), vertical raypaths are assumed for the near-surface layer. The figure
shows that surface features result in reduced traveltime
(or intercept time) profiles that have features at the same
spatial position (receiver location), regardless of the
recording direction.
Figure 5-44 is a similar display to show the effect of
subsurface features on reduced traveltimes for both forward and reverse directions of recording. The true
refractor velocity for the left side of the model (V2) is
used to compute the reduced traveltimes. In this case,
subsurface features give rise to reduced traveltime (or
intercept time) profiles that are spatially displaced
between the two directions of recording.
If the two plots in Figure 5-44 are shifted toward each
other, they each show the subsurface features at the correct spatial position, which corresponds to the location
of the emergent ray from the refractor, as shown by
point E in Figure 5-38b. The distance the plots should be
moved is the offset distance or displacement, which is the
horizontal distance from the emergent point on the
refractor to the surface receiver location. This is a function of the near-surface velocities and the refractor
velocity and depth (discussed further in Sections 5.6.5
and 5.6.6). The change in the refractor velocity (V2) on
the right side of the model in Figure 5-44 causes the two
plots to diverge from each other; this is because an incorrect velocity (V2) is used to compute the reduced traveltimes for this part of the line. I show in Section 5.6.6 how
this divergence can be used to compute the correct
refractor velocity.

5.6.3

ABC Method (Method of


Differences)

The ABC method is designed to compute weathering


layer times (Edge and Laby, 1931). The technique
requires data from a reversed profile with its reciprocal
time, and the data are analyzed at a common surface
location, such as in Figure 5-38a.
Raypaths for a reversed refraction profile are shown
in Figure 5-45. The traveltimes from a source at A to

189

V1

G
V2

Fig. 5-45. Raypaths for a reversed refraction profile to


illustrate the ABC method (method of differences).
Variables as in Figure 5-38.

receivers at B (tAB) and C (tAC) are given by


tAB =

AD DE EB
+
+
V1
V2 V1

and
tAC =

AD DG GC
+
+
.
V1
V2
V1

Similarly, for a source at C to a receiver at B,


tCB =

CG GF FB
+
+
.
V1
V2 V1

If locations A, B, and C lie on a straight line, these three


relationships can be combined as follows:
tAB + tCB tAC =

BE + BF EF

.
V1
V2

(5.66)

When V2 is much greater than V1 and if the refractor


is almost flat under the receiver at location B, the distance EF is almost zero, and BE is nearly equal to BF.
Under these conditions, equation (5.66) can be simplified to
twB = 0.5(tAB + tCB tAC),
(5.67)
where twB is the time through the weathered layer at
location B, which can be expressed as
twB BE/V1.
The ABC method can thus be summarized as follows:
the sum of the refraction arrival times to a receiver from
a source located on either side of the receiver minus the
reciprocal time is equal to twice the time through the
weathered layer at the receiver location. The depth to
the base of the weathered layer is calculated by multiplying the time through the weathered layer (twB) by the
near-surface velocity (V1).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

190

Static Corrections for Seismic Reflection Surveys

If the delay time concept (Section 5.6.2) is used, which


was proposed after publication of the ABC method,
equation (5.66) can be simplified to
tdB = 0.5(tAB + tCB tAC),

(5.68)

where tdB is the delay time at location B. Equation (5.64)


or (5.65) is then used to compute the depth to the refractor from the delay time.
The ABC method is normally used for analysis of discrete reversed refraction profiles. In the case of multifold
recording, a receiver location such as B records data
from many possible pairs of source locations A and C.
Thus, many estimates can be made of the weathered
layer time or delay time at the receiver location, using
equations (5.67) or (5.68), and the results averaged. For
nonsurface sources, the relevant uphole times should be
added to the observed refraction arrival times using
equations (5.38) or (5.42) prior to calculating the weathering time or delay time (Section 5.5.7.1). All arrival
times analyzed must relate to the same refractor for
equations (5.67) or (5.68) to be valid. This requirement
often restricts the number of receiver locations that can
be used to compute valid refractor information and
results in a partial depth profile between the two source
locations.
Salvatori and Walling (1937) patented a slightly different technique designed to compute differential times
between two receiver locations rather than a specific
value at one location, as in the ABC method. For the
refraction profile shown in Figure 5-45, with a source at
A and receivers at B and C, the differential time in the
weathered layer between the two receiver locations
(tBC) is given by

X
tBC = tAC tAB
cos c ,
V2

(5.69)

where X is the distance between receiver locations B


and C, V2 is the refractor velocity, tAB and tAC are the
times from source A to receivers B and C, and c is the
critical angle.
With an additional source located at K (to the right of
receiver C in Figure 5-45), another estimate can be made
of the differential time in the weathered layer between
receiver locations B and C:

X
tBC = tKC tKB +
cos c,
V2

(5.70)

where tKC and tKB are the times from the source at location K to receivers C and B. Equations (5.69) and (5.70)
can be combined to give

tBC = 0.5 (tAC tAB ) + (tKC tKB ) cos c . (5.71)

Table 5-7. Computation of Weathering Thickness Using


ABC Method on Refraction Data in Figure 5-39a.
Distance Source A Source C tAB + tCB tAC a Refractor
(m)
Times (ms) Times (ms)
(ms)
Depthb (m)
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115

2
13
17
20
23
27
30
32
36
38
42
45
48
51
54
57
59
61
64
68
71
74
75
78

79
75
72
69
65
64
61
57
54
52
48
45
43
40
37
34
31
26
22
20
18
16
12
2

c
9
10
10
9
12
12
10
11
11
11
11
12
12
12
12
11
8
7
9
10
11
8

2.1
2.4
2.4
2.1
2.8
2.8
2.4
2.6
2.6
2.6
2.6
2.8
2.8
2.8
2.8
2.6
1.9
1.7
2.1
2.4
2.6
1.9

a Reciprocal time (t ) = 79 ms.


AC
b Refractor depth computed using a near-surface velocity of 475 m/s.
c indicates no value because one of the arrivals was judged not to be on the
refractor.
Depth estimates using the intercept-time method are 2.3 m at a distance of 1 m
(source A) and 2.0 m at 116 m (source C).

Thus, differential times through the weathered layer


can be obtained for contiguous receiver locations from
the observed traveltimes and then integrated to form a
continuous profile along the line. This profile requires
calibration to convert the values to absolute times; this
low-frequency (long-wavelength) component can be
obtained with data from uphole surveys or from another refraction technique on several segments of the line.
With multifold data, individual estimates between
receiver locations from each source location can be averaged. Examples using this basic approach were documented by Leven and Taylor (1989). The ABCD method
of Bahorich et al. (1982) is a similar technique, modified
to include source-to-receiver offset information to allow
for its use on crooked-line recording. In the derivation of
equation (5.71), it is assumed that the distance between
receiver locations B and C (X) is the same for the two
source locations. Since this is not necessarily true in
crooked-line recording, equation (5.71) must be modi-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

191

(a)

(a)

SP A
0
Depth (m)

Depth (m)

SP A
0
1
2

(b)

1
2

1
2

(c)

600

Velocity (m/s)

Velocity (m/s)

600

3
(c)

0
Depth (m)

Depth (m)

SP C

3
(b)

SP C

300

300
0

0 20

2040

100
6080
80
4060
Distance along
spread along
(m) spread (m)
Distance

115100

115

Fig. 5-46. Interpretation of data shown in Figure 5-39a using the ABC method (includes values for the end points from
the intercept-time method): (a) refractor depth profile using constant near-surface velocity of 475 m/s (vertical exaggeration 10); (b) dipping refractor based on depths from the two end-points in part (a) (vertical exaggeration 10); (c) nearsurface velocity profile which ties depth profile in (b) with the observed refraction arrival times.

fied to include the two differential offset values and the


refractor velocity. As in the ABC method, many estimates of the differential times can be averaged with
multifold recording.
The data plotted in Figure 5-39a are used to illustrate
the ABC method. The original picked times, time computations, and depth values along the spread are given
in Table 5-7. The refractor depth profile, including values from the intercept-time method listed in Section
5.6.1, is shown in Figure 5-46a. I should stress that the
irregular depth profile shown assumes that the nearsurface velocity is constant at 475 m/s, a value estimated only at each end of the profile. Alternatively, a
smooth refractor depth profile can be postulated, which
implies that the near-surface velocity varies along the
recording spread on the basis that the refraction arrival
times are not changed. This is illustrated in Figure 5-46b,
where a linear dip is assumed between the two end
points and the implied near-surface velocity profile indicates a range of 330550 m/s.
Many other permutations and combinations of nearsurface velocity and refractor depth are possible as a
result of undersampling of the near-surface velocities
(see Section 5.7). In addition, the range of possible errors
must be considered; in Figure 5-46b, a change of 1 ms in

the time observed at a distance of 85 m along the spread


(8 ms) implies a change in the near-surface velocity from
485 to 555 m/s. Thus, if the depth profile in Figure 5-46b
is realistic, some of the near-surface velocity variation
may be due to minor timing errors in the refraction
arrivals.
The near-surface model of Figure 5-40 (derived from
the timedistance display in Figure 5-39b) showed that
refraction arrivals were observed only from a shallow
refractor from one direction of recording. Also, reversed
coverage data were obtained over about half the recording spread on the deeper refractor. Thus, this data set is
not suitable for interpretation by the ABC method, especially for the shallow refractor. The deeper refractor
could be mapped, however, and converted to depth
using an average of the two overburden velocities.

5.6.4

Hagedoorn (Plus-Minus) Method

The Hagedoorn or plus-minus method (Hagedoorn,


1959) is normally used for shallow refraction surveys or
for estimating weathered layer corrections. It is based on
the earlier graphical approach of Thornburgh (1930), in
which wavefronts were constructed at equal time inter-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

192
A

Static Corrections for Seismic Reflection Surveys


B

The expressions for tAR, tBR, and tAB can be used to


rewrite equation (5.72) as
T+ = 2 R +

V1

AR BR AB
+

.
V2
V2 V2

Hence,
T+ = 2R,
because

V2

AR + BR = AB,

Fig. 5-47. Raypaths for a reversed refraction profile to


illustrate the Hagedoorn or plus-minus method. Variables
as in Figure 5-38.

vals from the two shots of a reversed refraction profile.


The plus times are used to give the traveltime from the
surface to the refractor, while the refractor velocity is
estimated from the minus times. Data are analyzed at a
common surface location, as shown in Figure 5-38b.
Similar techniques have been reported by Hagiwara and
Omote (1939) and Hawkins (1961). The wavefront
approach proposed by Hagedoorn is described in
Section 5.6.8.4; the derived formulas are described
below and verified using a raypath approach. Worked
examples from Yemen are shown by van Overmeeren
(1987), and Cummings (1979) has described how the
technique can be used on a pocket calculator.
Figure 5-47 shows raypaths for a reversed refraction
profile from two source locations to a common receiver
location. Using equation (5.61), we can express the
refraction arrival times from a source at A to receivers at
R (tAR) and B (tAB) as
tAR = A + R +

AB
,
V2

T = tAR tBR.

(T+)

is defined as the sum of the travThe plus time


eltimes from two sources located on either side of a
receiver minus the reciprocal time, or
T+ = tAR + tBR tAB .

(5.72)

(5.74)

Using the above relationships between traveltime and


delay time, we can rewrite equation (5.74) as
AR
BR
B R
V2
V2
AR - BR
B ) +
V2
AR - ( AB - AR )
B ) +
V2
AB 2 AR
.
B )
+
V2
V2

T = A + R +
= ( A
= ( A
= ( A
Thus,

2X
,
V2

(5.75)

where K is a constant and X is the offset from source A


to receiver R, equal to distance AR in Figure 5-47. Thus,
if T is computed for receivers between the two source
locations and is then plotted against the source-toreceiver offset X, the refractor velocity can be estimated
from the slope of the display.
An alternative definition of the minus time adds the
reciprocal time to equation (5.74):
T = tAR tBR + tAB.

BR
.
V2

(5.73)

The delay time at the receiver location is thus equal to


half the plus time.
The minus time (T) is calculated by subtracting the
times from the two sources located on either side of a
receiver:

T = K +

where A, B, and R are the delay times at locations A,


B, and R. Similarly, for the refraction arrival time from a
source at B to a receiver at R (tBR),
tBR = B + R +

R = T+/2.

AR
,
V2

and
tAB = A + B +

and thus

(5.76)

Using the approach used in the derivation of equation


(5.75), we can rewrite this as
T = 2 A + 2

AR
V2

or

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

193

Table 5-8. Computation of Weathering Thickness and Refractor Velocity Using Hagedoorn (Plus-Minus) Method
on Refraction Data in Figure 5-39a.
Distance
(m)
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115

Source A
Times
(ms)

Source C
Times
(ms)

2
13
17
20
23
27
30
32
36
38
42
45
48
51
54
57
59
61
64
68
71
74
75
78

79
75
72
69
65
64
61
57
54
52
48
45
43
40
37
34
31
26
22
20
18
16
12
2

tAR + tCR tAC a


(T+)
(ms)
c
9
10
10
9
12
12
10
11
11
11
11
12
12
12
12
11
8
7
9
10
11
8

Refractor
Depthb
(m)

2.2
2.5
2.5
2.2
3.0
3.0
2.5
2.7
2.7
2.7
2.7
3.0
3.0
3.0
3.0
2.7
2.0
1.7
2.2
2.5
2.7
2.0

tAR tCR
(T)
(ms)

62
55
49
42
37
31
25
18
14
-6
0
5
11
17
23
28
35
42
48
53
58
63

a Reciprocal time (t ) = 79 ms.


AC
b Refractor depth computed using a near-surface velocity of 475 m/s.
c indicates no value because one of the arrivals was judged not to be on the refractor.

T = K +

2X
,
V2

where K is a constant. This expression is similar to


equation (5.75), except for a different constant.
As with the ABC method, the plus-minus method is
normally used for the analysis of discrete reversed
refraction profiles. For multifold recording, a receiver
location such as R records data from many possible
pairs of source locations A and B; thus, many estimates
can be made of the receiver locations plus time and the
results averaged.
The data shown in Figure 5-39a are used to illustrate
the plus-minus method. The original picked times, time
computations, and the depth values along the spread are
shown in Table 5-8. These values are essentially the same
as those shown for the ABC method (Table 5-7), although
there are some minor differences in the refractor depth.
The simple depth conversion approach, which assumes
vertical travel in the weathered layer, was used in the
ABC method (Table 5-7). The depth conversion for the
Hagedoorn method includes the cos term in equations

(5.64) or (5.65) (see Section 5.6.2), and thus minor differences in refractor depth occur. However, the cos term is
often used in the ABC method, which results in identical
depths to the refractor. The minus times are listed in
Table 5-8 and plotted in Figure 5-48a. This indicates a
refractor velocity of about 1680 m/s compared to values
of 1690 and 1640 m/s obtained from the timedistance
displays in Figure 5-39a.
I suggested in Section 5.6.1 that the data in Figure
5-39c were too complex for the intercept-time method;
these data are now interpreted using the plus-minus
method on the assumption that all values are associated
with a single refractor. This can be verified if the minus
times are aligned and indicate a single refractor velocity.
The original picked times, plus times, and minus times
are listed in Tables 5-9 and 5-10 for the two recording
spreads; the minus times are plotted in Figure 5-48b.
Except for points at or close to the ends of the recording
spread, the data are aligned and indicate velocities of
4000 m/s between source locations K and L and
4140 m/s between L and M. This suggests that the data
plotted in Figure 5-39c represent a two-layer situation

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

194

Static Corrections for Seismic Reflection Surveys

(a)
80

T- (ms)

40
1680 m/s
0
-40
-80
0

40

20

60
80
Distance along spread (m)

100

115

(b)
SP L

SP K
100

SP M

T- (ms)

50
4000 m/s

4140 m/s

0
-50

-100
0

200

100

100
300 (0)
Distance along spread (m)

300

200

Fig. 5-48. Velocity estimates using the minus times of the Hagedoorn or plus-minus method: (a) LVL survey data from
Figure 5-39a; (b) data from the two reversed profiles shown in Figure 5-39c.

Table 5-9. Computation of Weathering Thickness and


Refractor Velocity Using Hagedoorn (Plus-Minus) Method
on Refraction Data in Figure 5-39c
(Source Locations K and L).
Distance
(m)

Source K
Times
(ms)

12.5
37.5
62.5
87.5
112.5
137.5
162.5
187.5
212.5
237.5
262.5
287.5
a Reciprocal time (t

Source L t KR + t LR t KLa t KR t LR
Times
(T+)
(T)
(ms)
(ms)
(ms)

14
37
52
61
72
79
78
83
85
89
93
95

96
95
90
91
87
86
69
64
52
45
34
20

KL) = 96 ms.

14
36
46
56
63
69
51
51
41
38
31
19

82
58
38
30
15
7
9
19
33
44
59
75

Table 5-10. Computation of Weathering Thickness and


Refractor Velocity Using Hagedoorn (Plus-Minus) Method
on Refraction Data in Figure 5-39c
(Source Locations L and M).
Distance
(m)
12.5
37.5
62.5
87.5
112.5
137.5
162.5
187.5
212.5
237.5
262.5
287.5

Source L
Times
(ms)

Source M t LR + t MR t LMa t LR t MR
Times
(T+)
(T)
(ms)
(ms)
(ms)

21
41
42
56
67
65
81
81
106
110
105
89

a Reciprocal time (t

91
90
83
81
84
66
71
62
72
65
47
21

LM) = 90 ms.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

22
41
35
47
61
41
62
53
88
85
62
20

-70
-49
-41
-25
-17
-1
10
19
34
45
58
68

Chapter 5Refraction Surveys


(a)
SP K

Delay time (ms)

60
40

195

Fig. 5-49. Interpretation


of data shown in Figure
5-39c using the
Hagedoorn or plusminus method: (a) delay
time profile; (b) refractor
depth profile. Vertical
exaggeration 5.

20
0

(b)

Depth (m)

0
10
20
30
0

100

200

300 (0)
100
Distance along spread (m)

200

300

and that the variations seen are due to large near-surface


changes within the recording spread.
The magnitude of these variations are shown in
Tables 5-9 and 5-10, where the plus time (T+) is twice the
delay time at the receiver location (equation 5.73). If we
exclude the closest offset data at source locations K and
L (because these did not record the refracted arrivals),
the delay times vary by 19 and 34 ms within the two
recording spreads. Figure 5-49 shows the delay time
profile and a depth profile using the near-surface velocity estimates, which vary from 1000 m/s at source K to
600 m/s at sources L and M. The refractor varies in
depth from 6 to 30 m, with some changes in excess of
10 m across a 25-m group interval. In Section 5.6.3, I
indicated that this apparent depth variation might exist
or that the refractor depth profile might be smoother
than this, implying that the near-surface velocity
changed along the line. This refractor depth and nearsurface velocity ambiguity is discussed further in
Section 5.7. The implications of errors in the depth
model are considered in Section 5.7, along with the
resulting errors in the computed datum static corrections from this model.

interpretation technique introduced by Palmer (1980,


1981, 1986) extends this, so that both common-surface
and common-subsurface approaches can be accommodated. As with the ABC and plus-minus methods, this
generalized reciprocal method (GRM) also requires forward
and reverse refraction arrival times together with the
reciprocal time. The method is relatively insensitive to
dip angles up to 20 and can be used in the presence of
velocity gradients. In addition, it is capable of detecting
the presence of hidden layers and velocity inversions,
providing the refractor offset distance or displacement (horizontal distance from the emergent point on the refractor to the surface receiver location) can be measured
with sufficient accuracy.
The first step in the procedure involves the computation of the XY value, which is twice the refractor offset
distance, and the refractor velocity. This is achieved
using the velocity analysis function and generalized timedepth values. In the raypath diagram in Figure 5-50,
sources are located at A and B and receivers at X and Y.
The velocity analysis function (tv) is defined as

5.6.5

where tAY is the traveltime from source A to receiver Y,


tBX is the traveltime from source B to receiver X, and tAB
is the traveltime from source A to source B, or the reciprocal time.
The velocity analysis function is considered to apply
at location G, midway between the two receiver loca-

Generalized Reciprocal Method


(GRM)

The ABC and plus-minus methods described earlier


both analyze refraction arrival information at a common
surface location, such as in Figure 5-38a. A refraction

tv =

1
(tAY tBX + tAB ) ,
2

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

(5.77)

196

Static Corrections for Seismic Reflection Surveys

V1

G E

by applying static corrections to the data, prior to a


detailed analysis of the deeper data.
The ABC and plus-minus methods are based on a
common-surface-location approach, which is equivalent
to setting the distance XY to zero. If XY = 0, equation
(5.77) is identical to the minus time expression for the
Hagedoorn method, as defined by equation (5.76).
The XY value and the apparent refractor velocity for
the n-th layer (Vn) can now be used to derive the generalized time-depth (tG), the one-way traveltime to the
refractor, defined by

V2

Fig. 5-50. Raypaths for a reversed refraction profile to


illustrate the generalized reciprocal method (GRM).
Variables as in Figure 5-38.

tions at X and Y. The XY value is the distance between X


and Y; the optimum value occurs when X and Y are the
surface locations corresponding to the two emergent
rays from the same point on the refractor, that is, when
D, G, and E (Figure 5-50) are coincident. For each
receiver location on the reversed refraction profile, a
series of velocity analysis functions (tv) are derived
using different values of XY. At the optimum XY value,
the velocity analysis function plot is the simplest, and its
slope is equal to the inverse of the apparent refractor
velocity.
When analyzing near-surface refraction data, it is
likely that the XY value will have several components
because it is based on the refractor depth, refractor
velocity, and near-surface velocity. For very shallow
refractors, the XY value is often less than the group
interval used, such that the effective XY value is zero,
whereas a deeper refractor has a larger XY value, which
can be a significant distance. In addition, a near-surface
feature or anomaly, such as a local change in elevation,
will have an XY value of zero, or close to zero. Because
the XY value is a composite of these different components, it is likely to change along the line if the refractor
depth or the relevant velocities change or if there are
near-surface changes. In the presence of near-surface
anomalies, such as elevation changes along the line, any
analysis of the XY value from refraction arrivals is likely
to be dominated by the surface anomalies, which have
an XY value of zero.
Thus, the logical first step in the interpretation of a
relatively deep refractor is to estimate the static corrections associated with the near surface and remove these
from the data (see Section 5.5.7.5). The XY value of the
shallow refractor is now estimated on the corrected
data. If deeper refractors are also present, it may be necessary to remove the effects of the shallow layer, again

tG =

1
XY
tAY + tBX tAB +
.
2
Vn

(5.78)

For practical reasons, the generalized time-depth values


are often calculated for a range of XY values at the same
time as the velocity analysis functions, so that the final
term in equation (5.78), involving XY and Vn, can be
neglected; it is introduced later once these parameters
are known. When distance XY = 0, that is, X and Y refer
to the same receiver location, equation (5.78) becomes
equivalent to equation (5.72) (plus times in Hagedoorn
method) or to equations (5.67) and (5.68) (ABC method).
In the case of multifold recording, receiver locations X
and Y record data from many possible pairs of source
locations A and B. Many estimates can thus be made of
the generalized time-depth, as given by equation (5.78),
and the results averaged.
The optimum XY value was defined by Palmer (1986,
120) as follows:
The finite value for which the velocity analysis function shows a minimum of lateral velocity changes and
for which the time-depths show a maximum of detail. It
assumes that the data have been satisfactorily corrected
for near surface variations, i.e. elevation and weathering
corrections have been made. Furthermore it assumes
that there is some irregularity in the refractor. Unless
such an irregularity exists, neither the velocity analyses
nor the time-depths will show extrema and the
focussing concept cannot be employed.

Figure 5-51 shows timedistance displays for a twolayer model containing a short inclined ramp with a dip
of about 11.3 at the interface between the two layers.
The depth to the interface varies from 20 to 30 m, and
the velocities of the two layers are 1000 and 2000 m/s.
For the recording spreadlength of 295 m (60 groups with
a 5-m group interval), the reciprocal time is 191 ms and
the XY distance, computed with equation (5.9), ranges
from 23 to 35 m along the profile.
Figure 5-52 shows generalized velocity analysis functions for this model data, using a range of XY values
from 0 to 50 m in 10-m steps. These displays include

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

Fig. 5-51. Near-surface model


containing an inclined step
and derived timedistance
displays of a reversed refraction profile.

200

Time (ms)

197

100

Depth (m)

50

100

150
Distance (m)

200

250

295

0
20
40

1000 m/s

2000 m/s

information from the second arrivals shown in Figure


5-51 to extend the range of the reversed coverage data.
Because variations in this data set are small, the values
are reduced by the refractor velocity (2000 m/s) and
shown on an expanded time scale to give a better visual
display for subsequent analysis. The annotated times on
the y-axis are correct for XY = 0, but subsequent displays
are time shifted in increments of 2 ms to avoid overplotting of points for different XY values. Thus, the XY =
40 m display is time shifted by 8 ms. The slope of the displays represent differences in velocity from the reference
velocity of 2000 m/s. As stated earlier, the velocity
analysis function shows a minimum of lateral velocity
changes at the optimum XY value. Figure 5-52 indicates
an XY value of 2030 m at a distance along the profile of
about 100 m and a value of 3040 m at a distance of
about 160 m.
Figure 5-53 shows generalized time-depth displays
for the same range of XY values, in which the term
involving the XY value and the refractor velocity in
equation (5.78) is neglected. The maximum detail in the
time-depth displays for the optimum XY value (see definition above) is observed at an XY value of 20 m at a
distance of about 100 m along the profile and of 3040 m
at about 160 m. Thus, evaluation of both the velocity
analysis function and the generalized time-depth displays leads to estimates of XY close to the theoretical values of 23 m at a distance of 100 m along the profile and
of 35 m at 150 m. The small time differences seen in
Figures 5-52 and 5-53 indicate the need for accurate

refraction arrival times. Small timing errors, or errors


associated with the removal of near-surface anomalies,
are likely to make the task of picking the XY value much
harder, or in some cases, impossible.
In some situations, the XY value can be estimated
from the original timedistance displays. The displays
in Figure 5-51 show anomalous zones in the two directions of recording, but it is not obvious that these are
associated with the same subsurface feature. However,
if these arrival times are reduced using a velocity close
to the refractor velocity, the correlation becomes clearer.
For example, Figure 5-54a shows the data from Figure
5-51 reduced using a velocity of 2000 m/s, and Figure
5-54b shows the same data reduced using a velocity of
1800 m/s. The anomalous parts of both pairs of profiles
are displaced by about 15 m at a distance of 100 m along
the profile and by about 35 m at 160 m. However, the
similarity of the two profiles is easier to see with the
correct refractor velocity (Figure 5-54a) than with one
that is in error by 10% (Figure 5-54b). Thus, on this data
set, a reasonable estimate of the XY value can be
obtained with the reduced refraction arrival times.
Section 5.6.6 shows that this technique to estimate the
refractor offset distance is also used in delay time interpretation methods.
Other examples of velocity analysis functions and
generalized time-depth displays are shown in Palmer
(1980, 1981, 1986, 1990, 1991) and Lankston and
Lankston (1986). An accurate determination of the XY
value is often difficult, especially if the subsurface fea-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

198

Static Corrections for Seismic Reflection Surveys

Fig. 5-52. Generalized velocity analysis function displays


for the traveltime data shown
in Figure 5-51 reduced with a
velocity of 2000 m/s, using a
range of XY values from 0 to
50 m. Annotated times are
correct for XY = 0; other displays are time shifted by
increments of 2 ms.

Reduced velocity analysis function (ms)

30
XY = 50 m
26

XY = 40 m
XY = 30 m

22

XY = 20 m
XY = 10 m

18

XY = 0

14
0

200

100

295

Distance (m)
10

Generalized time-depth (ms)

Fig. 5-53. Generalized timedepth displays, using a range


of XY values from 0 to 50 m,
for the traveltime data shown
in Figure 5-51.

20

XY = 0
XY = 10 m
30

XY = 20 m
XY = 30 m
XY = 40 m
XY = 50 m

40
100

200

295

Distance (m)

tures on the refractor are small. Under these conditions,


the XY value can be computed if sufficient information
is available about the layers above the refractor, which
can be obtained from an uphole survey. The XY value
was defined for the two-layer case (Section 5.3.1) by
equation (5.9); a general form of this equation for multiple layers is given by
n

XY = 2

Vj

Z jG tan sin 1 Vn ,

mated XY value and refractor velocity as defined in


equation (5.78), is now converted to depth. It was shown
by Palmer (1980, 1981, 1986) that for plane layers, the
generalized time-depth was related to the layer thickness (ZjG) and depth conversion factor (Vjn) by
n 1 Z

Vjn ,

tG =

jG

(5.80)

j =1

(5.79)

j =1

where Vn is the refractor velocity, ZjG is the layer thickness, Vj is the layer velocity, j is the layer index, and n is
the number of layers. If the XY value cannot be estimated from the refraction arrival times, velocity analysis
functions, or time-depth displays, and if no information
is available about the layers above the refractor, an XY
value of zero can often be used as a default value.
The time-depth profile (tG), updated with the esti-

where Vjn is defined as


Vjn

(V

2
n

VjVn
Vj2

1/ 2

where Vj and Vn are estimated or apparent velocities.


For the two-layer case (Section 5.3.1), the above two
equations reduce to equation (5.8) with its velocities of
V1 and V2, except for the factor of 2 relating to one-way
versus two-way times. Palmer (1980, 1981, 1986) recom-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

Reduced traveltime (ms)

(a)

199

Fig. 5-54. Reduced traveltime displays of the


model data shown in
Figure 5-51: (a) reduction
velocity = 2000 m/s;
(b) reduction velocity =
1800 m/s.

60
50
40
30

(b)
Reduced traveltime (ms)

60

40

20
100

295

200
Distance (m)

mended that the layer thicknesses be measured perpendicular to the base of each layer to reduce the dip sensitivity of this approximation. The refractor surface is then
constructed using a migration approach, in which arcs
are drawn with radii equal to the thickness of the layer
and the tangent to these arcs defines the interface.
For the data shown in Figure 5-51, and for subsequent analyses in Figures 5-52 to 5-54, the velocities for
the two layers are 1000 and 2000 m/s. Equation (5.80)
can be simplified for this two-layer case to
tG = ZG/V12,
where V12 = 1155 m/s. The construction of a depth profile for this data is shown in Figure 5-55. Layer thicknesses are computed every 5 m along the profile using
the generalized time-depth values given in Figure 5-53,
modified by the appropriate XY distance and refractor
velocity term in equation (5.78). More detail is shown on
the dipping section of the profile in Figure 5-55b, which
is a 4-fold enlargement of the scale in Figure 5-55a.
An alternative method of converting the generalized
time-depth values to depth uses the concept of the average velocity from the surface down to the refractor. It
was shown by Palmer (1980, 1981, 1986) that the average
velocity (Vav) can be expressed as
Vav

Vn2 XY
=

XY + 2tGVn

1/ 2

(5.81)

The total thickness (ZG) of all layers is then given by


ZG =

tGVav

V
cos sin 1 av
Vn

(5.82)

Use of the average velocity concept requires that the


optimum value of XY be accurately computed. For the
data in Figure 5-51, an XY value of about 20 m is
obtained at a distance of 100 m along the profile, which
is associated with a time-depth value (tG) of 17.3 ms.
Using a refractor velocity (Vn) of 2000 m/s, equation
(5.81) yields an average velocity of 950 m/s, 5% too low;
the theoretical value for the XY distance of 23 m gives
the correct average velocity of 1000 m/s.
To illustrate the sensitivity of the average velocity
technique, Table 5-11 shows average velocities for a
range of XY values from 5 to 40 m for the model data set.
This shows that an error of 7 m in the XY value translates into an average velocity error of 10%. More discussion of the errors using this depth conversion approach
can be found in Palmer (1980).
Section 5.7 covers various points on velocity control
for the depth conversion of refraction data, both in terms
of near-surface and refractor velocities. It includes a discussion on problems associated with the undersampling
of near-surface velocities. This can result in a choice
between a slowly varying near-surface velocity and an
irregular refractor depth profile, or a smooth refractor
and an irregular near-surface velocity profile.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Static Corrections for Seismic Reflection Surveys

Fig. 5-55. Refractor depth


profile construction of
model data from Figure
5-51 using the generalized
time-depth values from
Figure 5-53: (a) total profile; (b) enlarged display
over the dipping section of
the profile.

(a)
Distance (m)
50

100

150

200

250

0
Depth (m)

200

20
40

(b)
100

Distance (m)
130
120

110

140

150

Depth (m)

10

20

30

40

Table 5-11. Relationship Between Average Velocity and


XY Value of GRM for Model Data in Figure 5-51.

XY Valuea
(m)

Average Velocity
from Surface Down
to Refractor (Vav)
(m/s)

5
10
15
20
23
25
30
35
40

520
710
850
950
1000
1030
1100
1160
1210

Values of XY should be computed from the depth


profile using equation (5.79) to check that the depth section is consistent with the traveltime data. For example,
if the observed value of XY is greater than that computed from the final depth profile, either an undetected
layer is present or the computed thickness or velocity of
one or more of the layers is too small. Thus, the technique is capable of detecting the presence of hidden layers and velocity inversions, providing the XY value can
be measured from the refraction information with sufficient accuracy (Palmer, 1980, 1981, 1986; Lankston,
1989). Other case histories of shallow refractors using
the generalized reciprocal method have been presented
by Hatherly and Neville (1986), Kilty et al. (1986),
Palmer (1990, 1991), and Cotten and Gochioco (1991).

a XY value defined by equation (5.81).

5.6.6

Gardner Method and Other Delay


Time Techniques

The Gardner method is a refraction interpretation


technique based on the delay time concept (Section
5.6.2). The separation of the intercept time into its two
component delay times (source and receiver) was
demonstrated for areal recording techniques by
Gardner (1939a, b) and in later publications for in-line
Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

Traveltime (ms)

150

201

Fig. 5-56. Timedistance


displays for ten source
locations from LVL surveys on a segment of
line with large near-surface variations.

100

50

0
Location

470

480

490

profile recording (Gardner, 1967; Barry, 1967). The


method is generally applicable to refractor dips of up to
about 10 and requires forward and reverse refraction
time profiles. Data are analyzed at a common subsurface
location, as illustrated in Figure 5-38b, and the method
is applicable to shallow or deep refractors. Other delay
time techniques were presented by Barthelmes (1946),
Pakister and Black (1957), Layat (1967), Hollingshead
and Slater (1979), Coppens (1985), and specifically for
nonreversed profiles by Wyrobek (1956) and Lawton
(1989). A recent case history from an overthrust area was
documented by Keliher and Bishop (1993).
Eight major steps or processes required to convert
refraction arrival times to a refractor depth profile are
detailed below and illustrated by two data examples.
Some of these steps are not required when the objective
of the refraction survey is to model the near surface.
Step 1. Near-surface corrections, including weathering and elevation corrections, are computed and applied
to the original refraction arrival times to refer the data to
a datum plane (see Section 5.5.7.5). If the objective of the
refraction survey is to estimate the weathering thickness
or to map a shallow refractor for near-surface control,
then these corrections are not required.
Step 2. The intercept times (or reduced traveltimes)
for both the forward and reversed profiles are computed for each source location and plotted at the relevant
receiver locations. As shown in equation (5.59), the intercept time (t0) is defined as
t0 = ti (Xi/Va),
where ti is the refraction arrival time at offset Xi and Va
is the refractor velocity. The refractor velocity should be
a reasonable estimate; the initial value is not critical and
is refined later in the interpretation procedure (step 6).
Two examples are used to illustrate the procedures,
both of which have been used earlier for other refraction
interpretation techniques. The data shown in Figure

500

510

520

5-39c, characterized by a highly variable near-surface


layer, are part of a larger data set. Timedistance displays of the refraction arrivals are shown in Figure 5-56
for the forward shots of 10 reversed profiles on this line,
where each profile consists of 12 receiver locations with
a group interval of 25 m. The data used in Figure 5-39c
were from source locations 468.5, 480.5, and 492.5. The
intercept times for the data in Figure 5-56 are shown in
Figure 5-57a, where a refractor velocity of 5000 m/s is
used to compute the intercept times; intercept times for
the reversed shots are shown in Figure 5-57b. The
assumed refractor velocity of 5000 m/s is probably too
high, on the basis that the velocity estimated by the
plus-minus method on a subset of the data is about
4100 m/s (Section 5.6.4). However, an updated refractor
velocity can be estimated from the intercept-time data
(demonstrated in step 6).
The second example uses the model data described in
Section 5.6.5 and shown in Figure 5-51. The intercept
times are given in Figure 5-57c, which uses an assumed
refractor velocity of 1600 m/s. This contrasts with the
correct refractor velocity of 2000 m/s.
Step 3. The offset distance, or the horizontal distance
between the emergent point on the refractor and the surface, is now estimated from the intercept-time profiles.
This is half the distance R1R2 in Figure 5-38b or half the
XY value of the generalized reciprocal method (Section
5.6.5). If subsurface features are present, they can be used
to estimate the offset distance directly from the intercepttime displays. For example, the reduced traveltime
model data in Figure 5-43 showed that the profiles were
aligned, indicating a zero-offset distance. This is true for
surface features and for a shallow refractor when the offset distance is less than the group interval. However, the
model data shown in Figure 5-44 where subsurface features were present indicated a displacement of the two
profiles.
These comments are similar to those made about the
XY value of the generalized reciprocal method in Section

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Static Corrections for Seismic Reflection Surveys


(a)
75
50
25
0

(b)
100
Reduced traveltime (ms)

Fig. 5-57. Reduced traveltime displays: (a) data


from Figure 5-56 (forward
shots) reduced using
velocity of 5000 m/s; (b)
reverse shots for the segment of line shown in
Figure 5-56 reduced
using velocity of 5000
m/s; (c) model data
shown in Figure 5-51
reduced with velocity of
1600 m/s.

Reduced traveltime (ms)

202

75
50
25
0

Location

470

480

490

500

510

520

(c)

Reduced traveltime (ms)

60

40

20

0
0

200

100

295

Distance (m)

5.6.5. In near-surface surveys, the offset distance has several components, from a value of zero for near-surface
anomalies to a value based on the depth of the deepest
refractor being mapped. To solve for these requires that
the near-surface anomalies be estimated first; static corrections (see step 1) are then applied to the data to
remove these anomalies, followed by an analysis of the
corrected data for deeper targets.
Features of the intercept-time profiles shown in
Figures 5-57a and b are broadly aligned in their plotted
surface positions, indicating that the main features are at
or close to the surface. Thus, the appropriate offset distance to use for this data set is 0 m. The intercept-time
profiles for the model data in Figure 5-57c show equivalent anomalous zones. These are displaced from each
other by about 20 m at a distance of 100 m along the pro-

file and by about 35 m at a distance of 160 m, that is, offset distances of 10.0 and 17.5 m.
The offset distance can also be computed from the
intercept-time profiles, providing an estimate of the
velocity down to the refractor is available. However, the
intercept time depends on the refractor velocity used,
and at this point in the interpretation procedure, the
velocity and intercept times are approximate. The intercept time represents the sum of a source and receiver
delay time, which means that as a first approximation,
the delay time is half the intercept time. If significant dip
is present, it is necessary to include this factor in the partitioning of the intercept time into its two delay times.
The offset distance (x) for a two-layer case can be
expressed in terms of the delay time (tdS) using equations (5.9) and (5.64):

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

Fig. 5-58. Reduced traveltime or intercept-time


display of data shown in
Figure 5-57c replotted in
the offset position: (a)
forward profile; (b) average intercept-time profile; (c) reverse profile.

Reduced traveltime (ms)

50
40
(a)
30
(b)

20

203

(c)

10
0

100

200

295

Distance (m)

x=

V1td S tan c
cos c

(5.83)

where V1 is the near-surface velocity and c is the critical angle. Similar expressions can be derived for the
multilayer case.
Step 4. The intercept-time profiles for each source
location are now replotted in the offset position; that is,
they are spatially shifted toward each source location by
the offset distance computed in step 3. This moves the
profiles so that the emergent points from the refractor
are aligned for the two directions of recording. When the
offset distance is less than a group interval (often the
case for shallow refraction surveys), this step is not
needed. Thus, step 4 is not required for the data shown
in Figures 5-57a and b. The model data in Figure 5-57c
are replotted in their offset positions in Figure 5-58,
showing that the anomalous zone is now closely aligned
for the two directions of recording. Offset distances of
10.0 and 17.5 m were used at the ends of the profile and
interpolated values between these two control points
from a distance of 100 to 160 m along the profile.
Step 5. The individual intercept-time profiles for
each source location are now combined to produce continuous intercept-time curves or profiles for each direction of recording. Normally, the first profile on the line is
used as a reference or starting point. This profile is
extended by suitably time-shifted intercept-time profiles
from successive source locations to the end of the line.
Overlapping profiles indicate the same relative dips,
providing the refractor remains the same. Intercept
times that indicate a different refractor must not be
included in this extension process. The appropriate time
shift is the average time difference between intercepttime profiles at common receiver locations, which is
equal to the differential source delay time. This process
thus requires that there are a reasonable number of common receiver locations from successive source locations
in which the intercept times refer to the same refractor.

A continuous profile is thus drawn for each direction of


recording; this profile is therefore parallel with the individual intercept-time profiles produced in step 4.
Irregularities not seen on both profiles at the same
location (at the emergent point from the refractor) are
sometimes smoothed. This assumes that they are due to
inadequate near-surface corrections or errors in the timing of the refraction arrivals. However, differences can
also be caused by other factors, such as diffraction energy in the case of rapid changes in refractor dip. An alternative approach, especially when there are many values
at each receiver, is to average the values at each location
after the profiles are time shifted by the differential
source delay time.
In the data shown in Figures 5-57a and b, some of the
near-offset data do not parallel the other offsets. These
offsets indicate a slower velocity that may be from the
direct arrival or a shallower refractor and are therefore
not included in the extension of the intercept-time profiles. As an example of the procedure, the difference
between the forward shots at locations 468.5 and 472.5
(Figure 5-57a) varies from 6 to 12 ms for common receiver locations 475 to 480. An average value of 8 ms is used
to make the continuous relative intercept-time profile in
Figure 5-59a. Thus, intercept times for source location
472.5 are adjusted by 8 ms to refer all the values to the
reference profile (source location 468.5); the values at
each receiver location are then averaged. This process is
repeated along the line to generate a continuous intercept-time curve for each direction of recording, as
shown in Figures 5-59a and c. These intercept-time
curves are converted from relative times to absolute
delay times in step 7.
Step 6. The divergence between the continuous
intercept-time curves for the two directions of recording
gives an indication of the refractor velocity error. The
average of the two curves represents changes in
absolute intercept time and hence gives dip information.
The divergence arises because the comparisons in one
profile are between short- and long-offset traces, where-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Static Corrections for Seismic Reflection Surveys

Fig. 5-59. Continuous


relative intercept-time
curves generated from
the reduced traveltime
displays shown in
Figures 5-57a and b:
(a) forward shots; (b)
average intercept-time
profile; (c) reverse
shots.

120
(a)
Reduced traveltime (ms)

204

80

(b)

40

(c)

Location

470

480

as the situation is opposite in the other profile. This can


be equated with the situation described in Section 5.5.5
with respect to split spread recording. The conversion of
the average intercept-time curve to absolute delay times
is described in step 7.
The corrected refractor velocity (Vc) is estimated from
the divergence between the continuous intercept-time
curves using the following relationship:
1
1
t
=

,
Vc Va 2 x

(5.84)

where t is the divergence between two consecutive


receiver locations for the two directions of recording, Va
is the apparent refractor velocity used to compute the
intercept times in step 2, and x is the group interval. The
negative sign in equation (5.84) is used if the forward
intercept-time profile decreases in time in the direction
of wave travel with respect to the average of the two
profiles. Since the refractor velocity can change along
the line, the divergence between the continuous intercept-time curves should be estimated at various locations along the line.
The continuous intercept-time profiles shown in
Figures 5-59a and c diverge, on average, by 2.12 ms
between receiver locations. The forward shots (Figure
5-59a) increase in time in the direction of wave travel (left
to right) with respect to the average continuous intercept-time curve (Figure 5-59b) and hence the sign in
equation (5.84) is positive. The apparent velocity used to
compute the intercept times is 5000 m/s, and the group
interval is 25 m. Using equation (5.84), this results in a
corrected refractor velocity of 4125 m/s. This is similar to
the value of 4100 m/s obtained with the Hagedoorn or
plus-minus method (Section 5.6.4) for part of the data set.
The intercept-time profiles for the model data shown
in Figures 5-58a and c diverge by 25 ms over a distance
of 100 m. The forward shot (Figure 5-58a) decreases in
time in the direction of travel (left to right) with respect
to the average continuous intercept-time curve (Figure

490

510

500

520

5-58b) and hence the sign in equation (5.84) is negative.


The apparent velocity used to compute the intercept
times is 1600 m/s. Using equation (5.84), this results in a
corrected refractor velocity of 2000 m/s, the correct
velocity for this model data.
Step 7. The average intercept-time profile computed
in step 6 is now calibrated and partitioned into its component delay times. This profile shows the correct
refractor dip as it is plotted in the offset position; that is,
where the rays enter and emerge from the refractor, as
illustrated by points B and C in Figure 5-42. The offset
position of a source and one of its receivers is now located on the average intercept-time profile, and the time
difference or dip () is estimated between these two
points.
At any one source location, refraction arrivals are
generally not picked at the critical distance (the sourceto-receiver offset corresponding to the offset distance),
other than as second arrivals. In shallow refraction surveys, these are unlikely to be recorded. As a direct consequence, the calibration procedure can rarely be used
for the first (or last) source location on the line. This is
not the case for subsequent source locations, as refraction arrivals from other source locations are used to
extend the average intercept-time profile.
The correct or updated intercept time (t0) for the
selected receiver location can now be computed. In
other words, the intercept time estimated in step 2 is
recomputed with the corrected refractor velocity
obtained in step 6. The delay times for the two selected
locations (td1 and td2) are then given by
td1 =

t0 +
,
2

(5.85)

td 2 =

t0 ,
2

(5.86)

and

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)

Fig. 5-60. Near-surface


profiles from segment of
line with large near-surface variations: (a) delay
time profile; generated
from calibrated intercepttime profile shown in
Figure 5-59b; (b) nearsurface profile showing
surface elevation, refractor depths, and near-surface velocities.

60
Delay time (ms)

205

40
20
0

(b)
120

Elevation (m)

695

580

80
830

450

635

40
1000

0
Location

470

480

490

where the larger of the two delay times equates with the
greater time on the average intercept-time curve.
This calibration procedure should be carried out for
other source locations along the line. Thus, time factors
are computed to adjust the average intercept-time profile to an absolute delay time profile. In theory, once one
point on the average intercept-time profile is calibrated,
all other points can be computed by simply time shifting
the profile to tie the calibrated value. It is normally best,
however, to compute several calibration points along
the line. An alternative calibration approach is to tie the
intercept times to control points established by uphole
surveys.
Calibration of the average intercept-time profile in
Figure 5-59b is described for the forward shots at locations 472.5 and 500.5. In this example, the offset distance
is zero, so the dip information is estimated at the source
and receiver locations because these are coincident with
their offset positions. Between source location 472.5 and
receiver location 484, the dip is 6 ms (Figure 5-59b). The
intercept time at receiver location 484 is 63 ms (Figure
5-57a); this reduces to 50 ms after making an adjustment
for the refractor velocity change from the assumed
5000 m/s to the corrected value of 4125 m/s. Thus,
using equations (5.85) and (5.86), the delay times at locations 472.5 and 484 are 28 and 22 ms, respectively. The
dip between source location 500.5 and receiver location
512 is 38 ms (Figure 5-59b). The intercept time at receiver location 512 is 58 ms, which reduces to 45 ms after the
velocity correction. This results in delay times at locations 500.5 and 512 of 4 and 42 ms, respectively.

500

510

520

In this example, calibrations were performed for all


possible source locations. The numerical time shift from
the intercept-time profile in Figure 5-59b to the calibrated delay time profile in Figure 5-60a varies from 19 to
23 ms, with an average value of 21 ms. The value of this
time shift is arbitrary because it depends on which profile is used as the reference in step 5.
For the model data in Figure 5-58, reversed coverage
and the average intercept-time profile do not quite
extend to the offset position of the source location at
either end of the profile, at distances of 10 and 277.5 m
along the profile. However, one-way coverage does
extend to these points. Without the second arrival data
(shown as dashed lines in Figure 5-58), only one-way
coverage can be obtained over a considerable distance.
Under these conditions, the average intercept-time profile can be extended on the assumption that the refractor velocity is constant or can be extrapolated from the
end of the reversed coverage data. The one-way intercept-time curve is thus modified by the interpreted
velocity variation away from the last point of reversed
coverage. Once this has been done, the procedure is the
same as that used in the previous example, except for
the nonzero-offset distance.
Step 8. The delay time profile is now converted to
depth using equation (5.64) for the two-layer case or a
similar expression for the multilayer case. For many surveys, uphole information is available at discrete locations along the line to act as calibration points; these
should give information on all the layers from the sur-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

206

Static Corrections for Seismic Reflection Surveys

1000

Velocity (m/s)
3000
2000

4000

5000

sions, providing the offset distance can be measured


from the refraction arrival information with sufficient
accuracy.

Depth (m)

5.6.7

100

200

k = 10

8 6

k=2

Fig. 5-61. Plot of velocity as a function of depth where


V(z) = Az1/k; values of k range from 2 to 10, with a constant velocity of 1000 m/s at a depth of 10 m.

face down to the refractor. Section 5.7 details various


methods of velocity control for the conversion of refraction times to depth and their limitations, as well as
comments on undersampling of the near-surface velocity field. I also discuss the implications of errors in the
depth model on any subsequently computed datum static corrections.
The delay time profile for the intercept-time curves
shown in Figure 5-59 (after the calibration described in
step 7) is displayed in Figure 5-60a. The surface elevation and refractor depth profiles are shown in Figure
5-60b, together with the average near-surface velocities.
The velocities are based on direct arrival information at
each source location and are used to convert the delay
times to depth. In this example, the near-surface velocity varies considerably along the line, with values varying by almost a factor of two in a distance of 200 m.
The offset distance should now be computed from
the depth profile using equation (5.9) for the two-layer
case (Section 5.3.1) or equation (5.79) for the multilayer
case (Section 5.6.5). This allows for a check to ensure that
the depth section is consistent with the traveltime data.
In both these equations, it is necessary to divide the calculated distance by two because they are designed to
give a surface offset distance. If the observed offset distance is greater than the value computed from the final
depth profile, it indicates that an undetected layer is present or that the computed thickness or velocity of one or
more of the layers is too small. It is thus possible to
detect the presence of hidden layers and velocity inver-

Blondeau Method

The Blondeau method is a refraction technique used


to estimate weathering corrections in areas where the
near-surface layers have an instantaneous velocity proportional to a power of the depth. The method is often
applicable to areas where the near-surface layers are
characterized by differential compaction between the
top and base of each layer, such as glacial drift. It is normally used to estimate the vertical traveltime from the
surface to a specific depth below the near-surface layers.
The theory was originally developed by E. Blondeau
in the late 1930s for the single-layer case with the source
at the surface (Dobrin, 1976; Dobrin and Savit, 1988).
Handley (1954) described the basic approach for both
single- and two-layer cases and referred not only to the
Blondeau method but also to a similar approach
described by Banta (1941). The extension to accommodate deep holes was published by Duska (1963), who
described a modification of the Blondeau method by
Swartz. The extension to the multilayer case with surface sources was described by Musgrave and Bratton
(1967). A further description of the technique can be
found in Gendzwill (1978). Examples of its use in glacial
drift were given by Patterson (1964).
The instantaneous velocity in the near surface is
assumed to be defined by equation (5.26) in Section
5.3.5:
V(z) = Az1/k,
where V(z) is the velocity at depth z, and A and k are
constants. A plot of velocity as a function of depth, using
this relationship, is shown in Figure 5-61 for k = 2, , 10,
where the velocity at a depth of 10 m is 1000 m/s for all
values of k. Other authors (e.g., Hollister, 1967; Banta,
1941) have based their work on a slight variation of this
velocitydepth relationship, namely,
V(z) = A(C + z)1/n,
where A, C, and n are constants.
The velocity characteristics of the near-surface layers
are derived from timedistance curves. This implies that
the refraction arrival times should be corrected for any
major high-spatial-frequency (short-wavelength)
changes in the near surface so that underlying velocity
trends can be established. In other words, static corrections (see Section 5.5.7.5) should be applied if necessary
to adjust the data to a datum close to the surface prior to
an analysis by the Blondeau method. Consequently, the
time corrections obtained by the Blondeau method are

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


x
S

depth z (point A on the raypath in Figure 5-62). The relationship between the maximum depth of penetration
(zm) and the surface offset (x) can be expressed as
zm =

207

zm

x
,
F

(5.88)

where
F = 2k
A

Fig. 5-62. Source-to-receiver raypath for a single layer,


where the velocity increases with depth, to illustrate the
Blondeau method. Variables: x is source-to-receiver offset, zm is maximum depth of penetration of raypath, z is
depth to a point on raypath, and is angle between raypath and vertical at depth z (after Musgrave and Bratton,
1967).

normally used at control locations along the line and not


for each source and receiver location.
It is often difficult to tell whether a timedistance
curve represents several discrete layers or a continuous
increase in velocity with depth, especially in the presence of noisy picks or time shifts due to improperly corrected near-surface anomalies. However, the two different situations are likely to lead to similar depth profiles
for the main refractor. This is because the multilayer
approach requires an average velocity from surface to
refractor because the intervening layers will be hidden
layers. An alternative approach to define the near-surface velocity structure in areas where velocity increases
with depth is with an inversion or tomographic
approach (see Sections 5.6.9.1 and 5.6.9.3).
I describe the Blondeau method here by defining and
listing the key equations that are required. This is followed by a step-by-step procedure to convert observed
traveltimes to vertical traveltimes for a single near-surface layer (Section 5.6.7.1) and for the two-layer case
(Section 5.6.7.2). Derivation of the equations that follow
(equations 5.875.94) are described by Duska (1963),
Bratton and Musgrave (1966), and Musgrave and
Bratton (1967).
Figure 5-62 shows a raypath from a source at S to a
receiver at R in a near-surface layer that has a velocity
increase with depth defined by equation (5.26). The
depth to any point on the raypath (z) can be expressed
in terms of the maximum depth of penetration of the
raypath (zm) as
z = zm sink ,

(5.87)

where is the angle between the raypath and vertical at

/2

= 0 sin

d .

(5.89)

The traveltime (T) from the surface back to the surface, or the time to travel the distance S to R in Figure
5-62, is defined by
G B
zm ,
A

(5.90)

B = 1 (1/k)

(5.91)

T=
where
and
G = 2k

/2

= 0 sin

k2

d .

(5.92)

The vertical time (tv) from the surface to a depth of zm is


defined in terms of the total traveltime (T) by
tv = T/F.

(5.93)

Thus, factor F represents both the ratio between the offset and maximum depth of penetration of the ray (equation 5.88) and the above relationship between vertical
time and total traveltime, (equation 5.93).
From equations (5.88) and (5.90), it follows that
log T = log G log A + B(log x log F).

(5.94)

Thus, the slope of the surface-to-surface arrival times (T)


plotted against distance (x) on log-log paper is equal to
factor B. If the source is in a hole below the surface, the
uphole time should be added to the observed times to
obtain surface-to-surface times. However, this simple
approach does not give the correct surface-to-surface
times, and the resulting analysis will indicate that the
acceleration with depth is higher than the correct value
(Duska, 1963). Figure 5-62 demonstrates this point for a
source located at A. If the uphole time is added to the
observed traveltimes, the raypath simulated is HAR;
this is quite different from the true surface-to-surface
path SAR through point A. Conversion factors that must
be applied to obtain compatible distances and traveltimes have been given by Duska (1963). Further comments on the treatment of uphole time are included in
step 6 of single-layer procedures in Section 5.6.7.1.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

208

Static Corrections for Seismic Reflection Surveys

20

250
(a)

16

200
(b)

12
Factor F

Time (ms)

150

100

50

0
0.2

0.4

0.6
Slope B

0.8

1.0

Fig. 5-63. Relationship between the slope B and the factor F of the Blondeau method (after Musgrave and
Bratton, 1967).

5.6.7.1 Single-Layer Method


The procedures required to convert observed traveltimes to a vertical traveltime for a single near-surface
layer using the Blondeau method are detailed below in
steps 1 to 6 and are illustrated with a data example.
Extension to the two-layer case is described in Section
5.6.7.2.
Step 1. Near-surface corrections are computed and
applied to the observed traveltimes to refer the data to a
smooth or flat datum close to the surface (see Section
5.5.7.5). These corrections are required only to adjust for
differences at receiver locations; they are generally not
necessary if only minor variations occur in the recording
spread. The uphole time is added to the arrival times,
although for deep holes, one should consider using the
more exact treatment described by Duska (1963) to compute the correct surface-to-surface times (see step 6).
Step 2. The surface-to-surface traveltimes are plotted
against the recorded distance on a logarithmic scale. As
indicated by equation (5.94), the slope of this plot is equal
to factor B. If more than one distinct slope is observed,
this indicates the presence of additional layers, with differing velocity depth relationships. The procedure for the
two-layer case is described in Section 5.6.7.2.

100

200

300

Distance (m)

Fig. 5-64. Timedistance displays from an area where the


velocity increases with depth: (a) source at the surface;
(b) source at a depth of 20 m.

The value of B is normally in the range of 0.50.9.


Equation (5.91) shows that this is equivalent to values of
k from 2 to 8. A B value of 0.5 (k = 2) implies that acceleration is constant, whereas a smaller value implies
increasing acceleration with depth. Higher values of k
are often treated using a constant-velocity approach. An
example of variations in velocity for different values of
k (from 2 to 10) was shown in Figure 5-61.
Step 3. The value of factor F is now computed from
slope B established in step 2. The relationship between
factors B and F was shown by Handley (1954) and by
Musgrave and Bratton (1967); tabulated values are plotted in Figure 5-63.
Step 4. The depth to the reference elevation or
datum is established. This is equal to the required maximum depth of penetration, or depth zm in Figure 5-62.
Equation (5.88) is then used to compute the horizontal
distance of the corresponding raypath (x).
Step 5. The surface-to-surface traveltime (T) corresponding to distance x (computed in step 4) is read from
the timedistance plot generated in step 2. The vertical
time to the reference elevation or datum (tv) is computed from traveltime T using equation (5.93).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

The timedistance display from a surface source


shown in Figure 5-64a indicates the presence of either
many thin layers or a continuous increase in velocity
with depth. The same subsurface model is used to generate the data in Figure 5-64b, but here the source is at a
depth of 20 m with an associated uphole time of 23 ms.
The inflection point in this display is due to the raypaths
for the first arrivals not penetrating below the depth of
the source at short offsets. A logarithmic display of the
surface source data is shown in Figure 5-65a, where
slope B equals 0.67. Similar plots of the deep-hole data
are shown in Figures 5-65b and c, with an uphole time
of 23 ms added to the observed traveltimes in the latter.
In both cases, the slope varies with offset but is nearly
constant at offsets >70 m for uncorrected data and about
100 m for corrected times. At these far offsets, slope B is
equal to 0.75 and 0.645 for the uncorrected and uphole
corrected data, respectively.
Table 5-12 lists various parameters required to estimate vertical times from the surface to depths of 20, 40,
and 60 m. This includes all three cases: surface source
data, uncorrected deep-hole data, and deep-hole data
adjusted by the uphole time. The correct times are those
listed for the surface source data plotted in Figure 5-65a.
The uncorrected deep-hole data indicate times that are
about 6 ms too small, whereas the uphole corrected

2.5

2.0
Log time in ms

Step 6. If the hole depth is greater than a few meters


and a simple time shift is used in step 1 to add the
uphole times to the arrival times, a correction factor
should be computed. This minimizes the error introduced in step 1, although it is not necessary if the full
correction described by Duska (1963) is used. The correction factor involves computing the vertical time (tv)
for a maximum depth of penetration (zm) equal to the
hole depth using the procedure described in steps 4 and
5. The difference between this pseudo-uphole time and
the observed uphole time is the correction factor to be
applied to the vertical traveltimes (tv) from step 5.

209

(c)
(a)
1.5

(b)

1.0
1.0

1.5
2.0
Log distance in m

2.5

Fig. 5-65. Timedistance display of the data shown in


Figure 5-64 plotted on a logarithmic scale: (a) source at
the surface; (b) source at a depth of 20 m; (c) source at a
depth of 20 m with uphole time (23 ms) added to the
arrival times.

times have a smaller error of 2 ms too large. The correction factor for the uphole time corrected deep-hole data
(step 6) is the difference between 25.6 ms, the time
equivalent of the 20-m shot depth, and the uphole time
of 23 ms. This means that a correction of 2.6 ms should
be applied to the times listed in Table 5-12. With the
addition of these corrections, the time error for the
uphole corrected times is reduced from 2 to 1 ms and the
adjusted times are now less than the correct value.

Table 5-12. Computation of Vertical Times from Surface to Depths of 20, 40, and 60 m
Using Blondeau Method on Data in Figure 5-65.
Source

TimeDistance
Display in

zm
(m)

x
(m)

T
(ms)

tv
(ms)

Surface

Figure 5-65a

0.670

4.02

Deep hole

Figure 5-65b

0.750

4.72

Deep holea

Figure 5-65c

0.645

3.86

20
40
60
20
40
60
20
40
60

80.4
160.8
241.2
94.4
188.8
283.2
77.7
154.4
231.6

94.0
150.0
197.0
87.5
147.5
198.5
99.0
150.0
194.5

23.4
37.3
49.0
18.5
31.3
42.1
25.6
38.9
50.4

a Uphole time of 23 ms added to arrival times.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

210

Static Corrections for Seismic Reflection Surveys

(a)
x
x

x1
S

R1

R2

z1

zm

Step 2. The surface-to-surface traveltimes are plotted against distance on a logarithmic scale and the two
slopes B1 and B2 are estimated from the display (see step
2 in Section 5.6.7.1). Figure 5-66a schematically shows
raypaths for the two-layer case from a source at S to two
receiver locations R1 and R2. These correspond to surface locations of rays from the source which have penetrated down to the interface between the two layers
(depth z1) and to the depth of the reference elevation or
datum (depth zm). The associated logarithmic display is
shown in Figure 5-66b. In most cases, near-surface conditions are such that B2 is greater than B1, as in Figure
5-66b, and thus, steps 38 are followed. If B2 is less than
B1, an alternative procedure is required, as outlined in
step 3 and steps 915.

(b)

Slope = B2

Step 3. The values of factors F1 and F2 are now estimated from the slope values B1 and B2 using the relationship plotted in Figure 5-63 (see step 3 in Section
5.6.7.1 for the single-layer case).

Log T

T1

Step 4. The distance to where the slope changes


from B1 to B2 (x1 in Figure 5-66b) is measured from the
logarithmic display in step 2. The thickness of the first
layer (z1) is now computed using x1 and F1 in equation
(5.88).

Slope = B1

x1

Log x

Fig. 5-66. Two-layer model where the velocities increase


with depth to illustrate the Blondeau method: (a) raypaths from a source to two receiver locations; these
correspond to rays penetrating to depths of z1 and zm;
(b) timedistance display plotted on a logarithmic scale.
Variables: x1 and x are source-to-receiver offsets for rays
penetrating to depths z1 and zm, T1 and T are corresponding surface-to-surface traveltimes, x = x x1, and
z = zm z1 (after Musgrave and Bratton, 1967).

5.6.7.2 Two-Layer Method


Section 5.6.7.1 described the procedure to convert
observed traveltimes to a vertical traveltime for a single
near-surface layer using the Blondeau method. When
two near-surface layers each have a velocity proportional to a power of the depth, a modified procedure is used
to evaluate the times from the surface to a specific depth
within the second layer. This procedure is described
below in steps 1 to 15.
Step 1. Near-surface corrections for the receiver
locations are computed and applied to the data. For
deep-hole data, the uphole time should also be added to
the arrival times. Further details about these options are
given in step 1 for the single-layer case (Section 5.6.7.1).

Step 5. The depth to the reference elevation or


datum is established. This corresponds to the required
maximum depth of penetration (zm) in Figure 5-66a. The
penetration in the second layer (z) is the difference
between z1 (from step 4) and zm. The distance or additional offset (x) between locations R1 and R2 in Figure
5-66a is computed using z and factor F2 in equation
(5.88). The total distance (x) from source S to receiver R2
is the sum of x1 and x.
Step 6. Surface-to-surface traveltimes T and T1 corresponding to distances x and x1 (steps 4 and 5) are read
from the timedistance display generated in step 2.
Step 7. The vertical time to the reference elevation or
datum (tv) is computed from traveltimes T and T1 using
a modified version of equation (5.93):
tv =

T1 T T1
+
.
F1
F2

(5.95)

Step 8. For deep-hole data, a correction factor is now


computed and applied using the procedure in step 6 for
the single-layer case (Section 5.6.7.1). If the hole depth is
less than z1 (depth to the interface between the two layers), only a single layer conversion procedure is required
to establish the equivalent or pseudo-uphole time.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Step 9. When slope B2 is less than B1 on the logarithmic display in step 2, the distance at which the slope
changes (xj) no longer directly corresponds to the depth
of the first layer (z1). This distance and the corresponding traveltime (Tj) are less than the offset and traveltime
values that correspond to the depth of the first layer,
defined as x1 and T1 in Figure 5-66b. This is because the
wavefronts from the top of the second layer arrive
before the wavefronts from the lower part of the top
layer.
Step 10. The thickness of the upper layer (z1) is estimated from a modified version of equation (5.88)
(Handley, 1954; Musgrave and Bratton, 1967):
z1 =

xj
F2

Step 11. The depth to the reference elevation or


datum is established. This corresponds to the required
maximum depth of penetration (zm) in Figure 5-66a.
The penetration in the second layer (z) is the difference between z1 (from step 10) and zm. The distance or
additional offset (x) is computed using z and factor
F2 in equation (5.88); the total distance (x) is the sum of
xj and x.
Step 12. The depth to the interface between the two
layers (z1 from step 10) is used with factor F1 in equation
(5.88) to compute distance x1. This distance corresponds
to a raypath with a maximum depth of penetration of z1.
Step 13. The surface-to-surface traveltimes T, T1,
and Tj corresponding to distances x, x1, and xj computed
in steps 11, 12, and 9 are read from the timedistance
display generated in step 2.
Step 14. The vertical time to the reference elevation
or datum (tv) is computed from traveltimes T, T1, and Tj
using a modified version of equation (5.93):
tv =

T1 T Tj
+
.
F1
F2

(5.96)

Step 15. For deep-hole data, a correction factor


should now be computed and applied to the vertical
times, as described in step 6 for the single-layer case
(Section 5.6.7.1) and step 8 above.

5.6.8

Wavefront Methods

Graphical wavefront techniques for the interpretation of refraction data were described by Ansel (1930),
Thornburgh (1930), Baumgarte (1955), Rockwell (1967),
and Schenck (1967). These techniques involve the construction of wavefronts associated with source and

211

receiver locations for two directions of recording and are


based on a common-subsurface-location approach, as
illustrated in Figure 5-38b. A raypath derivative was
described by Ackermann et al. (1982).
The general approach described here uses the arrival
times at various locations along the line, together with
the near-surface velocity, to construct a set of emerging
wavefronts. The refractor is defined where the sum of
the wavefront times from the two directions of recording is equal to the reciprocal time. The refractor velocity
is estimated from the distance between intersecting
wavefronts at the level of the refractor. The relief on the
refractor is thus dependent on the near-surface velocity
but is independent of the refractor velocity. The singlelayer case is described in Section 5.6.8.1.
To map deep refractors, shallower ones are mapped
first, then the velocity and depth of the shallower
refractor are used in the construction of wavefronts
through the upper layers. Deep refractors are again
defined using reciprocal time criteria. Section 5.6.8.2
describes the modifications required for the multilayer
situation.
Analytical raypath approaches proposed by Jones
and Jovanovich (1985) and Ak (1990) are briefly
described in Section 5.6.8.3. Hagedoorns (1959) plusminus method (described in Section 5.6.4) is based on
the Thornburgh approach and his paper included a
description of Thornburghs method. The plus-minus
method and how it relates to Thornburghs approach is
described in Section 5.6.8.4. Several authors have proposed that the graphical approach be implemented with
a numerical downward-continuation technique on the
recorded wavefield, or refraction arrival times, and this
is described in Section 5.6.8.5.

5.6.8.1 Single-Layer Method


The following steps are involved in constructing the
position of the refractor from refraction arrival times
using a general wavefront approach. The procedure is
illustrated with the two-layer model in Figure 5-51.
Step 1. Weathering corrections are computed and
applied to the observed refraction arrival times to
remove the time delays associated with the weathered
layer (see Section 5.5.7.5). An elevation correction is not
necessary, as the wavefront technique can be applied
from an irregular surface profile, such as the base of the
weathered layer. If the objective of the refraction survey
is to map the base of the weathered layer, no near-surface corrections are applied.
Step 2. Timedistance curves for the two recording
directions are plotted using the corrected arrival times.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

212

Static Corrections for Seismic Reflection Surveys

Fig. 5-67. Generation of


composite timedistance
curves where two refractors
are present.

Source 10

Time

Source 5

Source 1
location

Composite timedistance curves are now generated to


define a corrected arrival time, or pseudo-arrival time,
for as many locations as possible for each refractor. This
implies that a timedistance profile is synthesized from
the source locations along the line, such as the one in
Figure 5-12 (Section 5.4.1). This results in a set of arrival
times for each refractor and often represents data that
could not be recorded directly in the field. This is
because first arrivals are not observed at offsets less than
the crossover distance and deeper refractors are
observed on the longer offsets. As indicated in Section
5.4.1, the formation of a composite timedistance curve
involves the law of parallelism (Sjgren, 1980), also
referred to as a phantoming technique (Lankston and
Lankston, 1986).
If two or more refractors are present, a composite
timedistance curve is generated for each refractor
(Section 5.6.8.2). A two-layer example is shown in Figure
5-67 in which timedistance segments from individual
source locations are identified to show where they contribute to the composite timedistance curve. Source
location 1 is shown as the reference location, although
any of the other source locations could have been used.
The formation of the continuous relative intercept-time
curves of Figure 5-59 (from reduced time displays in
Figures 5-57a and b) is another example of a composite
timedistance curve, although generated on reduced
time displays.
For the model data shown in Figure 5-51, only one
reversed profile is present and therefore the data shown
are equivalent to a composite timedistance curve for
the line. The dashed lines represent secondary arrival
times that are used in step 3, along with the first arrivals
from the refractor, to determine the location of the emergent wavefronts.

8
7
Distance

10

Step 3. The emergent wavefronts for the shallowest


refractor are now constructed at equal time increments.
This requires the composite timedistance curves and
the velocity from the reference surface down to the
refractor. The time increment is chosen on the basis of
the complexity of the refractor surface; thus if long linear segments are observed on the timedistance curve, a
fairly large time increment should be satisfactory. The
times at which wavefronts are constructed are such that
the set of times from one direction of recording plus
those from the other direction can be combined to produce the reciprocal time. For the model data in Figure
5-51, the reciprocal time is 191 ms. Thus, times of 40, 50,
60 ms, and so on from one direction of recording and 41,
51, 61 ms, and so on from the other direction are complementary, as combinations such as 40 and 151 ms and
120 and 71 ms give total times of 191 ms.
The wavefront position at the reference surface is
established by finding the spatial location corresponding to its time on the composite timedistance curve. In
most cases, this requires interpolation between the plotted points. For example, for the forward profile of the
model data shown in Figure 5-51, where the source location is at the left end of the line, a wavefront time of 150
ms corresponds to a distance of 213 m along the profile.
For the reverse profile, a wavefront time of 101 ms corresponds to a distance of 197 m. The positions of the
wavefronts at the surface for the central part of this line
are shown in Figure 5-68 at times of 60150 ms for the
forward profile and 81151 ms for the reverse profile.
Positions of emergent wavefronts at earlier times are
now constructed at a distance of V1t from the surface
location, where V1 is the near-surface velocity and t is
the time increment. Specifically, arcs of radius nV1t
(where n = 1, 2, 3, . . . and t is the time increment

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)
50

Distance (m)
150

100

200

213

250

Depth (m)

10
20
140

30
40
40

50
(b)

50

100

80

60

250

200

150

100

120

Depth (m)

10
151

20
30
40

131

(c)

91

111

50

51

71

50

100

150

200

250

50

100

150

200

250

Depth (m)

10
20
30
40
50
(d)
0

Depth (m)

10

151

20
140

30
131

40
50

111

40

60

91

80

71

100

120

Fig. 5-68. Wavefront method; construction of emergent wavefronts and refractor depth profile for the model data shown
in Figure 5-51: (a) forward shots; (b) reverse shots; (c) forward and reverse shots; that is (a) and (b) combined; (d) as (c)
with construction arcs removed to highlight the wavefronts.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

214

Static Corrections for Seismic Reflection Surveys

between wavefronts) are drawn from each reference


surface location (the emergent wavefront position) back
toward the source location. Tangents to these arcs represent the emergent or directed wavefront. Thus, in Figure
5-68a, where the near-surface velocity V1 is 1000 m/s,
the location of the forward profile wavefront at a time of
100 ms is constructed from the following information:
a. Emergent point on the surface of the 100-ms
wavefront is at a distance of 127 m along the
profile.
b. Arc of radius (1 1000 0.010) m or 10 m is constructed centered on the emergent point of the
110-ms wavefront, at a distance of 142 m along
the profile.
c. Arc of radius (2 1000 0.010) m or 20 m is centered on the emergent point of the 120-ms wavefront, at a distance of 157 m along the profile.
d. Arc of radius (3 1000 0.010) m or 30 m is centered on the emergent point of the 130-ms wavefront, at a distance of 173 m along the profile.
e. Larger radii arcs are constructed centered on the
emergent points of the 140- and 150-ms wavefronts.
The process of arc construction, followed by the formation of tangents to these arcs, is carried out to define
the location of all emergent wavefronts down to the
depth of the refractor. These are shown in Figures 5-68a
and b for forward wavefront times of 40150 ms and
reversed times of 51151 ms. If the near-surface velocity
varies with depth, it is necessary to use a wavefront
chart, programmable calculator, or computer program
to determine the location of the emergent wavefronts
(e.g., Rockwell, 1967).
Step 4. The location of the refractor is defined at
points where the intersecting wavefronts have a total
time equal to the reciprocal time; the refractor velocity is
estimated from the distance between intersecting wavefronts at the refractor. For the data shown in Figure
5-68c, the dashed line joins eight wavefront intersections
where the total time equals the reciprocal time of 191 ms.
These range from a forward time of 50 ms and a
reversed time of 141 ms to times of 120 and 71 ms. The
depth profile defines a refractor that varies in depth
from 20 to 30 m and is identical to the original depth
model in Figure 5-51. The average distance between
these wavefronts is 20 m, equivalent to a refractor velocity of 2000 m/s for the wavefront increment of 10 ms.
At the ends of the line, wavefronts are defined only
for one direction of recording and therefore the refractor
cannot be defined by the above procedure. However, the
refractor definition can be extended outside the two-

way control, providing the refractor velocity is known


or can be extrapolated from the reversed coverage data.
Arcs of radius V2t, where V2 is the local refractor velocity, are constructed from the last defined point on the
refractor, at the intersection of two emergent wavefronts. The intersection of this arc with the next emergent wavefront defines the refractor location. This is
demonstrated in Figure 5-68d, where an arc of radius 20
m, based on an assumed refractor velocity of 2000 m/s,
is drawn centered on the location where the 120- and 71ms wavefronts intersect. The intersection of this arc with
the forward wavefront at 130 ms represents an extension
of the refractor depth profile.
If a different refractor velocity is postulated, such as
for a sensitivity check, it is a simple matter to test this by
drawing arcs with slightly different radii and observing
the change in the position of the refractor. Thus, the definition of the refractor depth profile only requires the
refractor velocity in areas having one-way coverage.
It is appropriate here to repeat a comment made in
the introduction to Section 5.6 on refraction interpretation techniques: the wavefront method gives the interpreter a good understanding of the refraction method
and some of its limitations. These include the portion of
the line where there is reversed coverage at the level of
the refractor and assumptions that must be made about
the refractor velocity to map the depth profile where
there is only one-way refraction control.

5.6.8.2 Multilayer Approach


The extension of the single-layer technique to the
multilayer approach is illustrated with a simple threelayer model that includes a dipping refractor. When
more than one refractor is to be mapped, the depth profile of the shallowest refractor is generated first using the
procedure indicated by steps 1 to 4 in Section 5.6.8.1. The
emergent wavefronts for the second refractor are now
constructed at equal time increments in a similar manner to that used for the first refractor. The wavefront
times from the two directions must again be such that
they can be combined to produce the reciprocal time.
The composite timedistance curve for the second
refractor is used to define the spatial location of the
emergent wavefronts at the reference surface.
The emergent wavefronts are first constructed
between the reference surface and the depth profile of
the shallowest refractor using the technique defined earlier. The intersection of these wavefronts with the shallow refractor depth profile defines the location of the
emergent wavefronts from the second refractor as if the
data had been recorded at the first or shallower refractor. This is equivalent to a downward-continuation pro-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)

215

150

Time (ms)

100

50

0
40

120

80

200

160

Distance (m)
(b)

Depth (m)

20

40
78
80

60

68

30 58

40

48

50

60

70

Fig. 5-69. Illustration of the wavefront method: (a) timedistance display for a three-layer model; (b) wavefront construction of data shown in (a) to illustrate the procedure for mapping a second interface.

cedure. These downward-continued locations are then


used to construct the emergent wavefronts in the second
layer. The procedure is the same as that used for the first
layer except for the velocity, which should be the interval velocity beneath the first refractor. This may be equal
to the refractor velocity, although in some cases it may
be appropriate to compensate for differences in velocities parallel versus perpendicular to the bedding planes
(see Sections 2.6.5 and 5.7.2.2). In some areas, the velocity may also be based on information from a deep
uphole survey.
Timedistance data for a three-layer example are
shown in Figure 5-69, in which the reciprocal time is 126
ms for the first layer and 108 ms for the second layer.
The shallowest refractor is at a depth of 15 m and has a
velocity of 2000 m/s. The construction of two wavefronts is shown in the figure, at a time of 50 ms in the forward direction with the source at a distance of 0 m and
at a time of 68 ms in the reverse direction. A velocity of

1000 m/s is used in the upper layer and 2000 m/s in the
second layer.
The refractor is defined at points where the intersecting wavefronts have a total time equal to the reciprocal
time for the second refractor. The refractor velocity and
the definition of the refractor away from the reversed
coverage data is estimated in the same way as that done
for the shallowest refractor. The position of the second
refractor is shown in Figure 5-69 at points where the
sum of the wavefront times from the two recording
directions is equal to the reciprocal time of 108 ms. The
position of the interface is extended past the end of the
reversed coverage, at a distance of 135 m, using the forward profile wavefronts and a refractor velocity of 3500
m/s. This velocity was estimated on the refractor where
there was two-way control. The construction is shown
with three different radii arcs centered on the intersection of the 60- and 48-ms wavefronts to define three successive wavefronts at 10-ms intervals.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Static Corrections for Seismic Reflection Surveys

SA

SB
R 1 R2

SC
R3 R4

tB6

R5 R 6

Traveltime

216

tB6
tcr

tB3
tB

V1

H
L

V2

R2

R3

R4

R5

R6

Distance from SP SB

Fig. 5-70. Selected raypaths from sources SA, SB, and SC


to receivers R1 to R6 to illustrate the construction of possible emergent points on the refractor. Variables: V1 is
near-surface velocity and V2 is refractor velocity (after
Jones and Jovanovich, 1985; Ak, 1990).

Deeper refractors are constructed in a similar manner; that is, by first mapping the shallowest refractor,
then using this information to define the next refractor,
and so on until all the shallower refractors are mapped.
This information is then used to trace the wavefronts
down to the target refractor so that it can be mapped.

5.6.8.3 Raypath Techniques


Adaptations of the wavefront method to a raypath
approach proposed by Jones and Jovanovich (1985) and
Ak (1990) trace the emergent ray back from the surface
using the emergent angle at the surface. The emergent
angle is estimated from the near-surface velocity, the
incremental arrival time between successive receiver
locations, and the receiver location spacing. This procedure is shown in Figure 5-70 in which selected raypaths
are drawn from three source locations to receiver locations R1 to R6. Both methods search for emergent points
on the refractor, such as H, which corresponds to receiver location R4.
The procedure proposed by Jones and Jovanovich
(1985) makes traveltime comparisons between two
receiver locations that straddle the critical distance from
the source location. For example, for the source at location SB, where the collocated entry and emergent point
on the refractor is H and the critical distance is SBR4, the
two receiver locations selected could be R3 and R6. The
refraction arrival time to receiver location R6 is defined
as tB6, which is the time for raypath SBHMR6. The equivalent time to receiver location R3 is tB3, which is com-

Fig. 5-71. Traveltimes from a source (SB) to two receiver


locations (R3 and R6) to illustrate the estimation of the
critical distance. Variables: tcr is traveltime for raypath
SBHR4 (Figure 5-70), tB6 for SBHMR6, tB3 for raypaths
(SBH + LR3 LH), t 'B3 for raypath SBGR3, and t 'B6 for
SBKR6 (after Jones and Jovanovich, 1985).

posed of the times for raypaths SBH and LR3 minus the
time for raypath LH. Time tB3 cannot be measured
directly from observations but can be computed using
the refraction arrival times from a more distant source
by back projection using the law of parallelism, as illustrated in Figures 5-12 and 5-67. This assumes that the
offset is such that the same refractor is involved. In the
case of a source at location SA, the relevant expression to
compute tB3 is
tB3 = tB6 tA6 + tA3 ,
where tA6 is the time from source SA to receiver R6 and
tA3 is the time from source SA to receiver R3.
In addition, the traveltimes t'B3 and t'B6 for raypaths
SBGR3 and SBKR6 are computed, assuming that the
entire distance traveled is at the near-surface velocity
(V1). These raypath distances are computed using the
emergent angle at the surface. Simple raypaths and the
appropriate velocity can be used to show that when
receiver locations R3 and R6 lie on either side of the critical distance from the source location,
tB6 < t'B6 ,
and
tB3 > t'B3 .
If these four values are plotted, they appear as in
Figure 5-71; the lines drawn between tB3 and tB6, and
between t'B3 and t'B6 intersect at the critical distance. The
surface location corresponding to the critical distance,
the near-surface velocity (V1), and the time for raypath
SBHR4 (tcr) are used to project the raypath down to the
collocated entry and emergent point on the refractor,
point H in Figure 5-70. This procedure is repeated for

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)

Source A
Surface

Source B

W
av
ef
so fromront
ur
s
ce
A

nts
fro
e
av m
W fro rce B
u
so

V1

V2

5.6.8.4 Hagedoorn Method

(b)
tA3
N
tA2
K
tB2

tA1
2kt

such as G is tried with relevant receivers R3 and R1; this


process is repeated until a match is obtained. This is
unlikely to occur with the original data due to the relatively coarse sampling between receiver locations, so
raypaths for additional points between the receiver locations are computed. Once the emergent point on the
refractor corresponding to receiver location R1 (point H
in Figure 5-70) is established, the process is repeated for
the other receiver locations along the line.

Refracted
wavefronts

tB3

217

V1t
tB1

Fig. 5-72. Wavefront diagram to illustrate the Hagedoorn


or plus-minus method: (a) wavefronts from sources A
and B including a short segment of the refracted wavefronts in the lower layer; wavefront spacing of t; (b)
enlargement of a few wavefronts from (a). Variables:
2kt is vertical separation of plus lines, c is critical
angle, tA1 to tA3 are wavefront times from source A, and
tB1 to tB3 are wavefront times from source B; others as
in Figure 5-70.

each source location (twice in the case of spilt recording)


to build up a profile of the refractor.
In the method proposed by Ak (1990), the refractor is
defined at points where the sum of the refraction arrival
times from the two recording directions equals the reciprocal time, as used in the wavefront method. This is
achieved on a trial-and-error basis by checking which of
the intersection points (such as those in Figure 5-70 for
the emergent rays at R1) correspond to a point on the
refractor. If point F is on the refractor, the sum of the
traveltimes for raypaths FR1 and FR2 is equal to the sum
of the refraction arrival times from source SA to receiver
R2, and from source SC to receiver R1, minus the reciprocal time between the two source locations. The near-surface velocity and raypath distances computed using the
emergent angle at the surface are used to compute the
traveltimes. If this is not true, the next intersection point

Hagedoorn (1959) considered the Thornburgh (1930)


wavefront method to be an excellent technique, but that
a short cut approach might be more appropriate in shallow surveys that lacked near-surface velocity information. A general description of the method based on the
wavefront approach is given here. A raypath verification
of the method was given in Section 5.6.4, which included data examples to illustrate the technique.
Figure 5-72a depicts emergent wavefronts from a
refractor for two recording directions, along with the
upper part of the wavefronts within the refractor. The
horizontal dashed lines and the vertical dotted lines represent the plus and minus times of Hagedoorn (discussed later). The location of a refractor was shown in
Section 5.6.8 to be at points where intersecting wavefronts from two directions of recording have a total time
equal to the reciprocal time. That is,
tA + tB = tAB ,
or
tA + tB tAB = 0,
where tA and tB are the refractor arrival times from
source locations A and B, and tAB is the reciprocal time.
The above equations thus define the interface shown in
Figure 5-72a.
Figure 5-72b shows an enlargement of one of the diamond-shaped features in Figure 5-72a, composed of two
pairs of intersecting emergent wavefronts with time separations of t. At points M and K, the two wavefront
times TM and TK are defined as
TM = tA1 + tB1 ,
and
TK = tA2 + tB2
= (tA1 + t) + (tB1 + t)
= TM + 2t .
If point M is on the refractor, it follows from the
above relationship and other intersecting wavefronts
that the horizontal dashed lines in Figure 5-72 are nt
above the refractor, where n = 2, 4, 6, . . . . These are the
plus lines and their vertical separation is the distance

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

218

Static Corrections for Seismic Reflection Surveys

KM or 2kt (Figure 5-72b). From the geometry in Figure


5-72b, we have

5.6.8.5 Downward-Continuation Procedure

tA2 tB3 = (tA1 + t) (tB1 + 2t)

The wavefront method is generally considered to be


a graphical approach, although several authors have
proposed that it be implemented using a numerical
downward-continuation technique on the recorded
wavefield or refraction arrival times (e.g., Taner, 1986;
Hill, 1987; Brckl, 1991; Aldridge and Oldenburg, 1992;
Taner et al., 1992; Qin et al., 1993; Brckl and Kohlbeck,
1994). These methods have the potential benefit of
allowing a much greater use of the wavefront technique
because the graphical approach is rarely used because of
the considerable time needed to construct the wavefronts. As in the graphical method, the reciprocal time is
picked manually; the near-surface velocity is estimated
from external information, such as direct arrivals and
uphole survey information.
Hill (1987) used a numerical downward continuation
in the tau-p domain. At the refractor, the sum of the forward and reverse times is equal to the reciprocal time;
that is, the forward time is equal to the reciprocal time
minus the reverse time. Data from a synthetic model are
shown in Figure 5-73, and the intersecting wavefronts in
Figures 5-73ac represent the location of the refractor.
The technique works on the premise that the product of
these two times is nonzero near the refractor, and the
refractor image is obtained by summing this product
over time, as in Figure 5-73d. If overlapping reversed
profiles are used, the resulting data can be summed to
give the final time profile. An extension to the multilayer case has been shown by Selvi (1990).
Taner et al. (1992) proposed a new method for implementing the ABC technique. The key relationship in the
ABC method (see Section 5.6.3) is of the form

= tA1 tB1 t.

twB = 0.5(tAB + tCB tAC),

V t
,
2 kt = 1
cos c
where c is the critical angle. Thus,
k=

V1
2 cos c

(5.97)

or
k=

V1V2

V22

V12

1/ 2

(5.98)

Expressions (5.97) and (5.98) are similar to equations


(5.64) and (5.65) in Section 5.6.2. The height of the surface above the refractor is thus factor k times the plus
time at a receiver location, or conversely, the refractor is
this distance (depth) below the surface. The plus time is
the sum of the refraction arrival times from two sources
located on either side of a receiver minus the reciprocal
time.
I stated in Section 5.6.8 that the refractor velocity was
estimated from the distance between intersecting wavefronts at the level of the refractor. Thus, the distance JL
in Figure 5-72b is equal to V2t, where V2 is the refractor velocity. The difference in the wavefront times at
point J for the two directions of recording is given by
tA1 tB2 = tA1 tB1 t.
At point N, the difference is

Thus, the differences are identical at these two points.


The vertical lines in Figure 5-72, such as the one joining
J and N, are characterized by a constant time difference
between the two emergent wavefront times and are
called minus lines.
The horizontal distance between minus lines was
shown above to be V2t. Their equivalent time separation is computed by finding the difference in the two
wavefront times at points J and L; at L, the difference is
given by
tA2 tB1 = (tA1 + t) tB1.
When compared with the difference at point J, this
shows that the separation between consecutive minus
lines is 2t. The minus time is the difference between the
refraction arrival times from the two sources located on
either side of a receiver. If these minus times are plotted
against distance (the source-to-receiver offset), the slope
indicates the refractor velocity.

where twB is the time in the weathered layer at location


B, tAB and tCB are the refraction arrival times to receiver
location B from sources A and C, and tAC is the reciprocal time (see equations 5.67 and 5.68).
Taner et al. (1992) noted that if two traces were convolved, the output contained the product of the two
amplitude spectra and the sum of the phase spectra.
Thus, if two traces contain refracted arrivals at times of
tAB and tCB, their convolved output indicates an arrival
time of tAB plus tCB. If the convolved output is then
shifted by the reciprocal time, the output represents
twice the time through the weathered layer. When this
procedure is used on several receiver locations, a seismic
section is produced, and if overlapping profiles are
recorded, the images can be stacked together. This section can then be converted to depth by a poststack
migration procedure. A similar approach can also be
used for a full wavefront method, in which the refracted
data are downward continued to a specific depth for the

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


a(t)

b(tr-t)

(a)

5.6.9

219

Inversion, Time-Term, and


Tomographic Methods

DEPTH

The methods summarized in this section adopt a


slightly more general and mathematical approach to
refraction interpretation. Generalized linear inversion is
described in Section 5.6.9.1, and the surface-consistent
decomposition of the refraction arrival times, often
called the time-term method, is covered in Section 5.6.9.2.
Tomography was introduced in Section 4.5 with specific
reference to crosshole (hole-to-hole) and hole-to-surface
tomography; its application to refraction interpretation
is described in Section 5.6.9.3.
These techniques are widely used for refraction interpretation of crooked-line and especially 3-D surveys.
These are discussed in Sections 5.6.10 and 5.6.11, where
I show that many of the other refraction interpretation
techniques are not appropriate for such surveys. Section
5.6.11 includes additional references on techniques
specifically applicable to 3-D surveys.

(b)

(c)

5.6.9.1 Inversion
0
(d)

3050

3050

3050

DEPTH

(d)

HORIZONTAL POSITION (m)

Fig. 5-73. Downward continuation of refraction data;


wavefronts from two directions of recording [a(t) and
b(tr t)], where the total time equals the reciprocal time
(tr): (a) wavefronts at times of 0.5tr and 0.5tr; (b) wavefronts at times of 0.6tr and 0.4tr; (c) wavefronts at times
of 0.7tr and 0.3tr; (d) refractor image obtained from the
downward-continued data. The (trough) model is shown
in (a), (b), and (c). Subscripts a and b refer to data from
the two source locations (Hill, 1987).

forward and reverse profiles. This depth is checked by


convolving the data to obtain the difference between the
sum of the forward and reverse downward-continued
arrival times and the reciprocal time, which is zero if the
depth chosen is the actual depth of the refractor.
Aldridge and Oldenburg (1992) downward continued the arrival times using a finite-difference algorithm.
They generated a succession of wavefronts in a manner
similar to that of the graphical method, such as in Figure
5-68; they referred to the approach as a wavefront reconstruction technique. The refractor is again defined at
points where intersecting wavefronts from forward and
reverse source locations have a total time equal to the
reciprocal time. The refractor velocity is estimated from
the distance between these intersecting wavefronts at
the level of the refractor.

To many readers, the topic of inversion is synonymous


with generalized linear inversion (GLI) because this inversion technique has been widely publicized, both in technical papers and in software packages. Inversion is,
however, a much broader topic and is defined by Sheriff
(1991) as follows:
Deriving from field data a model to describe the subsurface that is consistent with the data. Determining the
cause from observation of effects. . . . Inversion means
solving for a spatial distribution of parameters which
could have produced an observed set of measurements.

This definition indicates that all refraction interpretation


techniques discussed in this chapter are inversion methods. Zanzi (1990), for example, compared various refraction interpretation techniques, such as the generalized
reciprocal, plus-minus, and wavefront methods, under
the published title Inversion of Refracted Arrivals: A Few
Problems. There are many references to inversion; representative ones include Backus and Gilbert (1967),
Jackson (1972), Wiggins (1972), Wiggins et al. (1976),
Cooke and Schneider (1983), Lines and Treitel (1984),
Lines (1988), Russell (1988), and Santosa and Symes
(1989). Lines (1988) refers to Inversion of Geophysical Data
in SEGs Geophysics Reprints Series and includes many
references on inversion.
The objective of the generalized linear inversion
approach is to produce a near-surface model that best fits
the observed refraction arrival times (Hampson and
Russell, 1984a, b; Schneider and Kuo, 1985). This requires
a comparison between the observed refraction arrival
times with times obtained by forward modeling of a

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

220

Static Corrections for Seismic Reflection Surveys

Pick first-arrival
times from
refraction records

time variation corresponds to compensating changes in


velocity and thickness. The overall inversion procedure
is summarized in the flow diagram of Figure 5-74. Three
arrays of values are involved in the process: the
observed refraction arrival times (T), the near-surface
model (M), and the modeled refraction arrival times (R):

Generate nearsurface model


(initial guess)

T = (t1, t2, , tn),

Ray trace model


to compute first
arrival times

M = (m1, m2, , mn),


and
R = (r1, r2, , rn).

Compute time
differences
between model
and actual times

Small
enough?

Update
near-surface
model

No

Yes
Output final
near-surface
model

Fig. 5-74. Generalized linear inversion (GLI); procedure


for near-surface model generation using refraction
arrival information.

near-surface model. The resulting differences in arrival


time are then used to adjust, or update, the near-surface
model; the process is then iterated until the differences
reach an acceptable level or are unchanged between iterations. The updating of the model, the critical part of the
process, is based on generalized linear inversion, a technique widely used in the analysis of geophysical data.
Because of nonuniqueness, however, a good match
between the model and the observed data may not be the
only solution (e.g., Cooke and Schneider, 1983; Lines and
Treitel, 1984).
The initial model is based on the available information and should be a reasonable estimate. However,
extra care should be taken with definition of the nearsurface velocity because it is generally not updated in
the process. This is caused by the general lack of direct
arrival information required for independent estimates
of the velocity. There is an inherent ambiguity in the
approach because a small variation in the time through
the weathered layer can represent a change in velocity
and/or thickness. In addition, in some situations, no

The forward modeling process should be capable of


efficiently handling the complexity of the near-surface
model at the requisite accuracy; in many cases, a raytracing approach is used. The mechanism for updating
the model has been summarized by Hampson and
Russell (1984a, b) and Schneider and Kuo (1985); further
details are in Backus and Gilbert (1967), Wiggins et al.
(1976), and Lines and Treitel (1984). In some implementations of the technique, the layer thickness and velocity
profiles are smoothed to improve the stability of the
inversion. At the end of the process, the high-spatial-frequency (short-wavelength) components are computed
separately and added to the profiles from the inversion
process. Averaging the values at common surface locations (at collocated or closely located source and receiver positions) is another way of stabilizing the process.
With multifold data, the resulting equations are normally overdetermined; that is, there are more equations
than unknowns or independent variables, so that a
least-squares reduction approach is generally used.
The inversion approach can be modified to include
both refraction and reflection information (e.g., Berge
and Beskow, 1985; Scheffers, 1992). Landa et al. (1995)
proposed an inversion technique in which the coherency is measured on common source gathers along the
modeled arrival times. A nonlinear optimization technique is used to update the near-surface model between
iterations. Similar optimization techniques are used, for
example, in the analysis of residual static corrections
(Section 7.6).
The inversion of arrivals in media for which the
velocity increases with depth is routinely performed in
seismology. In this situation, the direct arrivals from the
source are refracted progressively as the depth increases, and the downward raypath is reversed so that the
ray returns to the surface (see Section 5.3.5). Thus, these
arrivals are often referred to as diving or turning waves.
Several authors have used a similar approach for nearsurface or shallow targets (e.g., Greenhalgh and King,
1980, with a discussion on this paper in Whiteley et al.,
1981; Gelius et al., 1984; Berge and Beskow, 1985; Rhl
and Lschen, 1990; Simmons and Bernitsas, 1994). In

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


many cases, the Wiechert-Herglotz integral or method is
used (Slichter, 1932). Static corrections should be applied
when necessary to a datum fairly close to the surface
prior to this inversion so that the timedistance curve
indicates only subsurface information. This is the same
procedure recommended for the Blondeau method
(Section 5.6.7), which applies to a specific type of velocity versus depth relationship. In some cases, however, it
is difficult to tell whether a timedistance curve represents several discrete layers or a continuous increase in
velocity with depth. This is especially difficult in the
presence of noisy picks or time shifts due to near-surface
anomalies.

(5.99)

where t12 is the time from a source at location 1 to a


receiver at location 2, d1 and d2 are the source and receiver delay times, X is the offset from the source to the
receiver, and Vr is the refractor velocity. Equation (5.99)
can be rearranged to evaluate the source delay time in
terms of the other variables:
d1 = t12 d2

X
.
Vr

1
= a + bx + cx 2 + dx 3 ,
Vr ( x)

(5.101)

where a, b, c, and d are constants and x is the distance


along the line. Equation (5.101) can be combined with
equation (5.99) by setting
x=

x2 + x1 ,
2

X = (x2 x1),

For the time-term or decomposition technique, the


refraction arrival times are separated into their componentsthe downgoing and upgoing segments and the
segment along the refractor (e.g., Scheidegger and
Willmore, 1957; Willmore and Bancroft, 1960; Paal,
1987). These components can be defined by equations
(5.3) and (5.61) in Sections 5.3.1 and 5.6.2, respectively.
Equation (5.61) can be written as
X
,
Vr

An alternative approach was suggested by Farrell


and Euwema (1984a, b) in which the refractor velocity
Vr(x) is approximated by an expansion based on quadratic or cubic powers of station numbers along the line:

and

5.6.9.2 Time-Term or Decomposition


Method

t12 = d1 + d2 +

221

(5.100)

A similar equation can be written to estimate the


receiver delay time. If refraction moveout or stepout is
applied to the data, the last term in equations (5.99) and
(5.100) becomes a residual moveout term. These equations can be simplified further if the residual moveout
term is assumed to be negligible or zero. These expressions are similar to those used for the surface-consistent
decomposition of time shifts into their component
source and receiver residual static corrections (discussed
in Section 7.4.1). In multifold recording, a large number
of equations are obtained, with many more equations
than unknowns. These can be solved, for example, with
a Gauss-Seidel or conjugate gradient technique (see
Section 7.4.2). Zanzi and Carlini (1991) suggested that
the wavenumber-offset domain should be used because
it is a more cost-effective approach to this inversion.

where X is the offset from source to receiver and x1 and


x2 are the in-line coordinates, or distance along the line,
for the source and receiver locations. Thus, the observed
refraction arrival time (t12) is composed of two delay
times and a velocity factor that depends on three or four
constants and the position along the line. The large
numbers of equations obtained with multifold recording are again solved with a least-squares technique.

5.6.9.3 Tomography
Tomography can be used to derive a near-surface
model from refraction arrival times from a suite of
source and receiver locations. Tomography is defined by
Sheriff (1991) as follows:
A method for finding the velocity and reflectivity distribution from a multitude of observations using combinations of source and receiver locations. Derived from
the Greek for section drawing. Space is divided into
cells and the data are expressed as line integrals along
raypaths through the cells. . . . Tomographic methods
include the algebraic reconstruction technique (ART), the
simultaneous reconstruction technique (SIRT), and GaussSeidel methods.

Representative works on tomography include Dines


and Lytle (1979), Devaney (1984), Worthington (1984),
Bishop et al. (1985), Ivansson (1985, 1986), Cutler (1987),
Nolet (1987), Wong et al. (1987), Stewart (1991), and Lo
and Inderwiesen (1994).
In the tomographic technique, the subsurface is
divided into cells and the objective is to estimate the
velocity of each cell. The raypath from a source to a
receiver location is composed of ray segments located in
different cells; the refraction arrival time is computed
from the individual distances traveled and the velocity
in each cell. Figure 5-75a shows a two-layer near-surface
model, in which the subsurface has been subdivided

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

222

Static Corrections for Seismic Reflection Surveys

(a)
S

(b)

Fig. 5-75. Near-surface model illustrating velocity cells


used in the tomographic imaging technique: (a) two-layer
model and a source-to-receiver raypath; (b) source-toreceiver raypaths in a single layer model where the
velocity increases with depth.

into cells, and a source-to-receiver raypath. Computed


or modeled refraction arrival times from an initial nearsurface model are compared to observed values. The
forward modeling process, which is often undertaken
with ray tracing, varies from a complete and accurate
approach to one in which straight rays are used from the
refractor to or from the surface, even when the ray crosses cells with differing velocities.
The differences between the modeled and observed
times are used to update the model, which is achieved
by perturbing the velocities in the cells. This iterative
procedure of forward modeling, measuring time differences, and updating the model is carried out until the
time differences are below a preset threshold. This normally occurs after a significant number of iterations; the
differences are generally minimized in a least-squares
sense. It is generally desirable to apply some form of
smoothing to the model to enhance the stability of the
process, and there is merit in keeping the model simple.
This approach is similar to the one used in the inversion
technique (Section 5.6.9.1).
A general requirement of tomographic inversion is
that there should be a large number of different raypaths
going through each of the cells with a wide angle of coverage. However, unlike crosshole tomography discussed in Section 4.5, the number of different raypaths
for 2-D refraction recording is generally small. One factor is that raypaths are common at the level of the refractor from a source to several receiver locations. In areas
where the near-surface velocity increases with depth, as
a result of compaction, for example, the presence of the

diving waves (turning waves) ensures that there are a


large number of different raypaths. In this situation,
more velocity cells can be defined and their velocities
estimated beneath any one surface location, as shown in
Figure 5-75b. Raypath coverage can be increased if
reflected data are also observed from the interface,
although these are often difficult to pick because they
arrive after the first arrival. If the data are obtained, they
can be incorporated in the inversion process (e.g.,
Vesnaver and Boehm, 1994).
Various tomographic techniques using refraction
and/or direct arrival information for definition of the
near surface are documented in the literature (e.g.,
Chon and Dillon, 1986; de Amorim et al., 1987; Zhu and
McMechan, 1988; Olsen, 1989; White, 1989; White and
Milkereit, 1990; Docherty, 1992; Simmons and Backus,
1992; Zhu et al., 1992; Adams et al., 1994; Bell et al.,
1994; Macrides and Dennis, 1994). Qin et al. (1993) recommended the integration of a tomographic approach
with a wavefront construction technique (discussed in
Section 5.6.8.5) to improve the final model. Other
papers dealing specifically with 3-D surveys are listed
in Section 5.6.11.
Olsen (1989) suggested that the horizontal dimension of each cell should be the estimated horizontal projection of the slanted raypath from the refractor to the
surface. This is half the critical distance or half the XY
value of the generalized reciprocal method (GRM)
(Section 5.6.5), as given by equation (5.79). In practice,
however, it is typically equal to one or two receiver
group intervals. Cells are usually rectangular, except for
variations due to topography of the surface or refractor,
and have a constant velocity. An alternative approach is
to define the velocity by nodes on a rectangular grid;
this allows for subdivision into a triangular grid that
allows a constant velocity gradient across the subcell
(e.g., White, 1989; White and Milkereit, 1990).
As in the inversion process (Section 5.6.9.1), it is not
possible to establish whether a small time variation
through the weathered layer is due to a change in velocity or thickness or a combination of the two. In addition,
in some situations, no time variation will correspond to
compensating changes in velocity and thickness. Under
these circumstances, it is normally necessary to keep one
of these two variables fixed at the values specified by the
input model and not update it during the iterative
process. If, however, there are sufficient short-offset
traces, the direct arrivals can be used to assess the
changes in the near-surface velocity on an independent
basis. The direct arrivals may represent a constantvelocity near-surface layer or diving waves as a result of
a velocity increase with depth. In the latter case, an
improved near-surface velocity field can be obtained
(White, 1989; Simmons and Backus, 1992; Zhu et al.,

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)
B

A
R

V1

V2
(b)
A

(c)
A

B
R

Fig. 5-76. Near-surface model to illustrate differences


between straight- and crooked-line recording on refraction interpretation methods: (a) depth model with refracted raypaths from two sources to a receiver location;
(b) plan view of (a) for straight-line recording; (c) plan
view of (a) for crooked-line recording. Variables as in
Figure 5-70.

1992; Stefani, 1993). To achieve this improvement, it is


necessary to have a good distribution of rays through
each velocity cell.
An alternative approach for multifold data was suggested by Docherty (1992) which involved the generation of a continuous timedistance curve for the whole
line. The objective is to simulate a single-source location
for the line, resulting in a smaller matrix for the inversion process. This requires the computation of the average differential times between successive receiver locations. The law of parallelism implies that these times are
independent of the offset and source location, providing
the same refractor is involved (see Section 5.4.1 and
Figures 5-12 and 5-67).

5.6.10 Crooked-Line Surveys


Many of the methods described in earlier sections are
broadly limited to 2-D surveys because they make the
general assumption that two source locations, with one
or more receiver locations between them, lie on a

223

straight line. This is illustrated in Figures 5-76a and b,


which show a near-surface model and plan view of the
recording layout. Under these conditions, the travel
along the refractor from each source to the common
receiver location is duplicated by the travel between the
two source locations, and this commonality can be
exploited to evaluate delay times at the receiver location. This applies, for example, to the ABC method, the
plus-minus method, the generalized reciprocal method
(GRM), and the wavefront approach.
The plan view of an equivalent model with crookedline geometry is shown in Figure 5-76c. Here, the travel
path along the refractor between the two source locations is no longer the same as that from each source to
the common receiver location, and the simple relationship used in the straight-line case no longer applies. Any
comparison between two recorded traces may therefore
need to accommodate variations in refractor velocity
with azimuth or changes in refractor depth. The lack of
raypath commonality was also discussed in Section 5.5.5
on the crosscorrelation of refraction arrivals. There I
showed that this was also an issue for straight-line
recording when source locations were offset from the
receiver line (see Figure 5-26).
In addition, for both crooked-line recording and offset source locations, the raypaths through the near-surface layer are no longer identical at any one source or
receiver location. Thus, the delay time may not be constant for all refraction arrivals at a location. The value
will differ if there are local changes at the refractor or in
the near-surface velocity. This can also occur on regular
in-line recording, specifically for a split spread configuration, when there may be differences in the delay time
at a location from the two recording directions. These
differences are usually very small and are not normally
considered in the refraction analysis. However, they can
be computed by averaging the delay times at a location
as a function of the azimuth of the upcoming or downgoing rays or wavefronts. If these corrections are
required, they are normally estimated as residual static
corrections, usually after an average value has been
determined for the location (see Section 7.9).
Refraction interpretation methods applicable to
crooked-line recording must recognize and use the offsets of the individual traces and must not have a requirement for common raypaths along the refractor. Several
approaches can be used, although these normally do not
accommodate the variations with azimuth and other
factors such as local changes at the location (referred to
above). One of these approaches is the delay time
method (Section 5.6.6), which can be used for shallow
refractors, although it is not as easy to correct for refractor velocity errors as it is when the group interval is constant. The ABCD method (see Section 5.6.3) is appropri-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

224

Static Corrections for Seismic Reflection Surveys

(a)

SB
R10
R8
SA

R2

R4

R6

100 m
(b)
60

= 90

T+ (ms)

= 80
= 70

40

= 60
= 50
= 40
= 20
20
SA

R2

R4

R6

R8

R10

SB

Fig. 5-77. Illustration of the plus-minus refraction interpretation method on data acquired on a crooked line: (a)
short spread recording configuration (sources at SA and
SB, receivers at R1 to R11); (b) plus times (T+) for nearsurface model (constant velocities and refractor depth)
using geometry defined in (a) for line bend (angle )
varying from 0 to 90 (method assumes straight-line
recording).

ate since offset information is used. The Blondeau


method (Section 5.6.7) is also applicable, although the
resulting velocity function represents data that apply
beneath an area rather than below the source and receiver line. In addition, the inversion method, surface-consistent decomposition of arrival times, and tomographic
approach described in Section 5.6.9 are readily adaptable to crooked-line recording, although problems can
occur with nonuniform coverage of data. This point is
covered in more detail in the discussion of 3-D surveys
in Section 5.6.11.
A simple model shown in Figure 5-77 is used to
demonstrate when crooked-line data can be analyzed
using a straight-line assumption. This specific example
gives an idea of the limitations of the approach, but its
results are not intended to be used as a hard and fast
rule for all recording geometries and near-surface conditions. The group interval of the recording arrange-

ment in Figure 5-77a is 20 m and the short spreadlength


is appropriate for an LVL or weathering survey. The
near-surface and refractor velocities are 1000 and
2000 m/s, and the delay time at all locations is 10 ms.
The resulting refraction arrival times are interpreted
using the plus-minus method; the plus times are plotted
in Figure 5-77b with the angle between the two halves of
the spread varying from 0 to 90 in 10 increments. This
display demonstrates that, for this simple case, a small
bend in the spread (<20) can be tolerated, but that for
larger angles, the magnitude of the time error can be significant. For example, with a 40 bend, the plus times
vary from 20.6 to 27.2 ms along the spread instead of the
correct value of 20 ms, and for a 90 bend, the variation
is 23.2 to 55.1 ms.
If a delay time approach is used for crooked-line
recording (Section 5.6.6), the intercept times or reduced
traveltimes can be computed with the correct offset and
appropriate refractor velocity and plotted at the relevant
receiver location. Computation and application of the
offset distance (steps 3 and 4 in Section 5.6.6) to the
reduced traveltimes, which is designed to align the subsurface emergent point from two directions of recording, is no longer a simple graphical task. This is due to
the variable distance between successive receiver locations from a common source location. However, this is
not a problem for shallow anomalies in which the offset
distance is zero (equivalent to XY = 0 in the GRM
method). Construction of the continuous intercept-time
curve or profile (step 5, Section 5.6.6) may be difficult if
the refractor velocity is significantly in error. This is
because the differential times between consecutive
receiver locations for different source locations are likely to be different, again due to the variable distance
between the two receiver locations. This also implies
that if the refractor velocity is wrong, the correctional
factor can no longer be obtained simply by inspecting
the divergence of the forward and reverse curves.
These points are illustrated in Figure 5-78, which uses
the same model as in Figure 5-77 except that there are 24
receivers per source location and several source locations. Timedistance curves for seven source locations
are shown in Figure 5-78b for the situation when the line
bend is 60. These appear to indicate a change in the
refractor velocity at receiver location 25, which corresponds to the bend in the line; this is due to the sourceto-receiver offsets no longer increasing by the receiver
group interval but by a smaller distance. If reduced traveltimes are computed for these data, with the correct
refractor velocity of 2000 m/s, all values indicate a constant time of 20 ms.
When an incorrect refractor velocity is used, however, this is not the case and the display is no longer simple. This is illustrated in Figure 5-78c, which uses a

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)

Source
locations

R49

R40

R30
R2

R10

225

R20

(1,1)

(2,1)

(1,2)

(2,2)

(1,3)

(2,3)

(1,4)

(2,4)

Receiver
line
A

S1 S5 S9 S13 S17 S21 S25

(b)

Time (ms)

300
(1,9)

200

Fig. 5-79. A 3-D land recording arrangement using four


receiver lines (A to D); source locations specified by
their x, y coordinates; y coordinates of receiver lines are
0.5, 3.5, 6.5, and 9.5.

100
0

(2,9)

S1 S5 S9 S13 S17 S21 S25

Reduced traveltime (ms)

(c)

20
0
-20
-40
-60
-80

-100
Location 1

10

20

30

40

50

Fig. 5-78. Illustration of the delay time refraction interpretation technique on crooked-line data: (a) recording
layout (sources at S1 to S25, receivers at R2 to R49); (b)
timedistance displays for near-surface model (constant
velocities and refractor depth) using geometry defined in
(a); (c) reduced traveltime display of the traveltimes in (b)
using refractor velocity of 1600 m/s (correct velocity =
2000 m/s), including a continuous relative intercept-time
curve (dashed line).

refractor velocity of 1600 m/s (an error of 20%) and the


reduced traveltimes show anomalous values at receiver
locations above number 25. The time separation
between successive receiver locations varies for different source locations even though the same refractor is
involved. This is again due to the variable effective
group interval as a result of the line bend. A continuous
relative intercept-time curve for these data, using the
median difference between reduced traveltime curves
for successive receiver locations, is shown in Figure
5-78c as an extension of the curve for the first source
location on the line. The average slope of the line is
about 2.3 ms per group interval, slightly less than the
correct value of 2.5 ms for this model. If two recording
directions are available, the divergence of the curves can
be used to estimate the true refractor velocity.

Thus, crooked-line data can be interpreted as 2-D


data, although in practice, situations will occur in which
the complexity of the field geometry (specifically, bends
in the line) will cause suboptimum results. However, it
should be remembered that this approach does make
the assumption of a 2-D subsurface that has no lateral
changes in the velocity or depth of the refractor. In practice, these factors are likely to be a problem only in areas
characterized by rapid changes in near-surface layers.

5.6.11 3-D Surveys


Section 5.6.10 (on crooked-line surveys) addressed
some potential refraction interpretation problems that
occur when source and receiver locations are not on a
straight line. This was illustrated by simple recording
situations in Figure 5-76. The main issues were the noncommonality of raypaths along the refractor and of the
raypaths through the near-surface layers at source and
receiver locations. Other aspects were possible changes
in refractor velocity with azimuth and the dip of the
refractor in the cross-line direction.
Further complications arise with the interpretation of
3-D refraction surveys. The interpretation of refraction
data in marine surveys, including 3-D surveys, is
described in Section 5.6.12. In 3-D land recording, the
data can be acquired in many ways. A general approach
is illustrated in Figure 5-79, which shows a succession of
sources recorded by several widely spaced parallel
receiver lines. The next series of shots are obtained by
moving the receiver locations along the line in the normal roll-along manner, with the line of source locations
moved up an equivalent distance. In this recording
arrangement, a subset of the source locations can be

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Static Corrections for Seismic Reflection Surveys

used with a specific receiver line so that the resulting


data set is 2-D.
Thus, for the recording arrangement in Figure 5-79,
source locations (1,1), (2,1), are recorded by receiver
line A, source locations (1,2), (2,2), are also recorded by
receiver line A, source locations (1,3), (2,3), are recorded by receiver line B, and so on. With regular recording
geometry, it is generally possible to extract a subset of the
3-D recording to simulate 2-D recording, which is linear
acquisition geometry. Such an approach cannot include
all acquired data, but it is certainly capable of generating
a reasonable near-surface model for many surveys. If the
recording geometry is more complicated, with less well
organized receiver lines and source locations, this
approach is unlikely to be practical. An alternative
approach, which can use all refraction data for almost
any recording configuration, is an inversion, surface-consistent decomposition, or tomographic technique (see
Section 5.6.9). Specific applications and examples of 3-D
surface-consistent decomposition and inversion have
been presented by Baixas and Dupont (1988), Dent
(1989), and Stresau et al. (1992a, b), among others. Chon
and Dillon (1986), Kirchheimer (1988), Taner et al. (1988),
Brzostowski and McMechan (1992), Docherty and
Kappius (1993), Macrides and Dennis (1994), and others
have described 3-D tomography.
A major factor that must be comprehended in any
inversion or tomographic technique on a 3-D data set is
the likelihood of large changes of fold across the area
covered; this will normally be at its lowest at the edges
of the survey area. It is also necessary for these techniques to be able to deal with nonregular geometry since
this is often used on 3-D surveys. One of the key parameters required by these techniques is the time difference between the observed refraction arrival times and
those obtained by forward modeling of the current nearsurface model, which are subsequently used to update
the model. In multifold recording, many such observations are usually associated with any one source or
receiver location. As the surface fold decreases, there is
an associated reduction in the number of observations;
timing errors then become more of an issue, as they are
harder to identify or average out with the reduced statistics available. In the extreme case when there is only
one observation that happens to correspond to a noisy
trace and is thus likely to have a larger than normal time
error, this can lead to instability in the updating process.
Similar variations in fold can occur in the subsurface
domain; this affects the tomographic approach (Section
5.6.9.3) because the subsurface is typically divided into
rectangular refractor velocity cells (see Figure 5-80).
These potential problems associated with areas of
low fold suggest the need for smoothing of the near-surface model to stop it from being corrupted by a few wild

R1

y coordinates

226

R2

R3

R4
S
x coordinates

Fig. 5-80. Plan view of the near surface, with refractor


segments of raypaths from a source (S) to four receivers
(R1 to R4), to illustrate velocity cells used in 3-D tomographic imaging. Solid line is travel within refractor; dotted line is travel through weathered layer.

picks. Other analysis techniques can be used to minimize the problems caused by bad picks, such as the use
of a weighting factor based on pick reliability and a
median approach to perform any averaging (e.g.,
Kirchheimer, 1988). In some cases, it may be preferable
to exclude individual time picks if they are too far away
from a smooth or simple model. In addition, both the
inversion and tomographic approaches should produce
a more robust solution if a slowly varying near-surface
model is invoked.
If the tomographic technique is used, situations can
occur in which some cells do not have any rays going
through them and are therefore not updated by the
process. This must be taken into account in any subsequent smoothing operation, such as the use of a weighting scheme. In most cases, zero-fold cells associated
with surface locations are due to the refraction arrivals
not being picked; this may be as a result of excessive
noise or that the source-to-receiver offset is outside the
chosen range. The overall reliability of the layer thickness is based on the surface fold; the refractor velocity
requires both subsurface fold and a range of offsets.
The operators used to smooth the near-surface model
must comprehend its 3-D nature; bicubic splines, which
define surfaces, are one such approach. The operator
length is typically about one spreadlength or more for
the refractor velocity, and about half a spreadlength or
less for the layer thickness or near-surface velocity. To
compensate for the variation in fold, and hence reliability of the model information, some type of weighting
scheme is normally included in this operation. Any
remaining high-frequency (short-wavelength) compo-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


nents, especially of the layer thicknesses or near-surface
velocities, may then be estimated at the end of the
process, for example, by evaluating a set of surface-consistent delay times.
The raypaths from a source to four receiver locations
are shown in Figure 5-80, which also illustrates the
refractor velocity cells for the tomographic process. The
solid lines represent travel within the refractor and the
dotted lines travel through the weathered layer. This is
further illustration of the point made in Section 5.6.10
that the raypaths through the near-surface layer are no
longer identical at any one source or receiver location for
nonstraight-line recording. These differences can be
addressed at this stage by averaging the surface-consistent delay times over a range of azimuths, although they
are more likely to be evaluated as azimuth-dependent
residual static corrections (discussed in Section 7.9). An
alternative approach is to include the problem in the
inversion or tomographic process by using smaller cells,
for example. However, this is unlikely to be practical in
most cases because it leads to a significant reduction in
fold, meaning that additional smoothing is required and
that the values are no longer independent estimates.

5.6.12 Marine Surveys


Marine surveys are normally considered to be much
simpler than land surveys, both in terms of datum static corrections and recording geometry. For example, all
the lines are straight, with the major complexity being
the possibility of a multistreamer or multivessel operation. However, in some 3-D marine surveys, obstructions such as drilling platforms may cause source locations close to the obstruction to be less well organized.
Refraction arrivals from interfaces below the water
bottom are often recorded, but in many cases these
occur only as first arrivals at long offsets. Section 5.3.1
and Table 5-2 showed that the crossover distance for a
simple two-layer case could be expressed in terms of the
velocity ratio between the two layers and the depth to
the interface. For example, if the subwater-bottom material has a velocity of 2100 m/s (about 1.4 times the water
velocity) the crossover distance is 4.9 times the water
depth. This means that offsets in excess of 2500 m are
required for a water depth of 500 m. If the target refractor is below the first subwater-bottom layer, the required
offset is even larger. The critical distances are typically
larger than those for a land survey due to the smaller
velocity contrast between the weathered and subweathered layers and hence a larger critical angle. Thus, the
use of refraction first arrivals is generally limited to shallow water depths and refractors that are not too deep
(e.g., Meeder et al., 1988; Bell et al., 1994). It is therefore
advisable to do simple modeling to find the offset range

227

Common receiver
location
(in-line coordinates)

Sail line

Fig. 5-81. Plan view of a feathered marine streamer to


illustrate lack of common receiver locations in marine
recording.

where a specific refractor is recorded as a first arrival


prior to any detailed data analysis.
One major difference as compared to land surveys is
that the weathered-layer velocity (water layer) is well
defined (see Section 2.6.3, Table 2-4, and Figure 2-14). In
addition, the water depth under the sail line can normally be obtained from fathometer information. As indicated in Section 3.6, however, in deep water or with significant dip, the water-bottom profile should be migrated to obtain the true water depth below the vessel. I also
described in Section 3.6 the problems and some possible
solutions if the water bottom dips on a cross-line basis.
In the analysis of refraction arrivals for both 2-D and
3-D surveys, an additional problem is caused by streamer feathering. This is illustrated in Figure 5-81, where the
receiver locations are offset to one side of the sail line. If
streamer feathering is present, there are virtually no
common receiver locations, that is, locations with exactly the same coordinates. If, for example, the feathering
angle is almost constant, a far-offset trace and a shortoffset trace having the same sail line or in-line coordinate will have different cross-line coordinates. There
may also be a small variation in the in-line coordinate as
a result of minor changes in the source interval.
The cross-line effect can be quite large; for example, at
distances of 250 and 3250 m along a cable feathered at 3,
the receiver locations are 13 and 170 m from the line, a
difference of 157 m. If the feathering angle is 6, the difference increases to more than 300 m. These are approximate cross-line distances because they are computed
assuming the streamer is straight, which may not be the
case. These differences can thus be much larger, even
with modest feathering angles, than some of the issues
discussed in Section 5.6.10, including slight differences
in the raypath through the weathered layer due to
crooked-line recording and source locations offset from
the line (see Figures 5-26, 5-76, and 5-80).
Another raypath aspect that must be included in the
analysis is the location of the receiver with respect to the

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

228

Static Corrections for Seismic Reflection Surveys

entry or emergent point on the water bottom. This offset


is dependent on the critical angle and the water depth.
The offset can therefore be significant for deeper water
surveys and where the velocity contrast at the water bottom is comparatively small, and hence, a large critical
angle is present. Consequently, an anomaly on the water
bottom will be observed at a receiver location which can
be offset from the anomaly by several group intervals. If
the model is subsequently used to compute datum static corrections, vertical raypaths are assumed, which can
be very different from the refraction raypaths. In an area
with a highly irregular water-bottom profile, the vertical
raypath assumption is likely to be very poor. This means
that a wave-equation technique, or the derivation of
model-based ray-traced time shifts are required to
accommodate the raypaths (see Sections 6.2.3 and 6.2.4).
In both these solutions, a near-surface depth model is
required. The refraction data must therefore be migrated to correctly position any water-bottom anomalies
that are present (Lourantos et al., 1991). This is also true
for some anomalies on land surveys, but the distances
involved are generally much smaller and are often dominated by anomalies close to the surface where the
migration effect is small.
To accommodate the variable cross-line distances and
the lack of common receiver locations, surface binning
can be used in much the same way as subsurface binning is routinely performed in crooked-line and 3-D surveys (see Section 3.6). If possible, the size of the bin
should be determined by the wavelengths of the anticipated near-surface anomalies rather than simply using
the group interval. Thus, if the anomalies are fairly low
frequency (long wavelength), relatively large bins can
be used. In any analysis of the refraction data, however,
it is necessary to use the correct source-to-receiver offset
and not the distance from the source to the bin center.
The binning approach may be degraded as a result of
variable fold, especially if the bins are small in size,
unless some form of fold harmonization (such as
dynamic binning) is used. The size of the surface bins
must also be considered in the evaluation of residual
static corrections (see Chapter 7). In the decomposition
or time-term approach (Section 5.6.9.2), an alternative
technique is to estimate the delay times on a regular grid
using a gridding algorithm such as the moving averages
technique (Dent, 1990). In this technique, a weighted
average is computed using weights proportional to the
distance of the receiver to the grid point.
A case history of a 3-D survey in the Mississippi Delta
was described by Heldreth and OConnell (1993). The
map of the datum static correction values, derived from
a surface-consistent decomposition of the refracted
arrival times, was shown to have a good correlation
with the seafloor morphology.

5.6.13 Applicability and Choice of


Refraction Interpretation Methods
This chapter has described several well-known
refraction interpretation techniques and has included
citations to several other methods; many more techniques have been published over the years. With so
many interpretation methods available, the newcomer
is often at a loss to know where to start and probably
notices that experienced colleagues often have their own
preferred methods. Also, a newly introduced technique
is often considered to be the one to use. So the question
is this: does a correct or preferred method exist? The
answer is generally no because the appropriate method
normally depends on the near-surface complexity and
on how the data were acquired. In this chapter, I have
described some of the strengths and weaknesses of each
interpretation technique so that their applicability to a
specific data set can be determined.
However, the inherent limitations of the refraction
method must still be considered, such as the inability to
detect thin layers or velocity inversions and the difficulty of accurately picking refraction arrival times. In addition, there is the near-surface velocity and refractor
depth ambiguity issue mentioned earlier, in which the
interpreter must decide between a smooth refractor and
an irregular or erratic near-surface velocity profile, or
the converse (see Section 5.7.3.1). Thus, adequate QC is
a basic requirement. In some cases, this should include
forward modeling of the final depth model to ensure
that the times are close to the observed times, although
agreement normally means that the model is one of
many possible models.
The acquisition of refraction data was described in
Section 5.4. I showed that these data can be recorded at
discrete locations along the line in weathering or lowvelocity layer (LVL) surveys, which normally use a
short recording spread. Alternatively, the refraction
arrivals recorded as part of the reflection survey can be
used. Various refraction interpretation techniques
appropriate for these data acquisition methods are
described below. This is followed by a review of nearsurface topographic variations (described in Chapter 2)
and how these impact the choice of refraction interpretation techniques. Other factors that may affect the
interpretation techniques include whether the refractors involved are shallow or deep and whether the
velocity is nearly constant for any one layer or increases continuously with depth. In addition, the surface
redundancy of the refraction arrival times can be a factor. Crooked-line and 3-D surveys represent special
cases, and appropriate interpretation techniques for
these (including variations with azimuth) were
described in Sections 5.6.10 and 5.6.11.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


With most refraction interpretations, some interpretive judgment is required, and it is generally not possible to obtain a unique result. This is often due to insufficient data from undersampling of the near-surface
velocity along the line. For example, Ackermann et al.
(1986) showed timedistance curves that could be
explained by many possible near-surface models.
Correct refractor segment identification is important; for
example, an observed change in apparent velocity may
be due to a refractor change or to a change in dip.
Diffracted arrivals from faults can lead to misidentification of refractor segments; several examples were
shown by Hawkins and Whiteley (1982) and Goulty and
Brabham (1984).
The interpretation of LVL surveys, which are generally recorded at discrete locations along the line, is normally undertaken by the intercept-time method (Section
5.6.1), the ABC method (Section 5.6.3), the plus-minus
method (Section 5.6.4), or the generalized reciprocal
method (Section 5.6.5). In some areas where the nearsurface velocity increases with depth, the Blondeau
method (Section 5.6.7) or a similar approach is used. The
traditional approach is the intercept-time method in
which the interpretation is carried out on the timedistance curves. This approach is practical, providing the
data are recorded along a straight line and any high-frequency (short-wavelength) time shifts are minimal. This
is because of the assumption of a planar refractor (see
Section 5.6.1).
The computation of the time through the weathered
layer, or more correctly, the delay time associated with a
receiver location, can be measured by several techniques. These include the plus time of Hagedoorns
plus-minus method, the generalized time-depths of the
generalized reciprocal method (GRM), or the weathered
layer times of the ABC method with the cosine term
included. For near-surface anomalies, where XY = 0 for
the GRM approach, all these techniques and equations
are identical. Also, under these conditions, the refractor
velocity estimates using the minus times of the plusminus method are identical to the velocity analysis function of the GRM.
The near-surface velocity is generally estimated from
the direct arrival information, either as a constant velocity or as an increase in velocity with depth (see Sections
5.3.5 and 5.7.1.2). Uphole survey information (see
Section 5.7.1.1 and Chapter 4) should also be used when
available. Two refractor depth estimates are generally
obtained, one at either end of the reversed spread. In
some cases, it is appropriate to average these, and the
resulting averaged values of velocity and depth are normally assumed to apply at the mid-point of the recording spread.

229

Refraction interpretation using arrival times from a


typical LVL survey with a short recording spread is
almost independent of near-surface conditions and
topography. This is because the refractors mapped are
shallow and have an offset distance (or XY value for the
GRM) of zero. However, most of the methods have a
dip limitation that may become an issue for steeply
dipping refractors. However, if there are significant
near-surface changes in the recording spread that generate high-frequency (short-wavelength) static shifts on
the timedistance curves, the intercept-time method is
no longer applicable. This is also likely to be the case for
the Blondeau method, unless the anomalous arrival
times can be removed satisfactorily with a set of static
corrections (see Sections 5.5.7.5 and 5.6.7). However, the
other methods mentioned above are appropriate. In
areas with highly irregular weathered layers, continuous coverage data is usually required if the near-surface
information is to be used later to compute datum static
corrections. For economic reasons, this information is
often obtained from the refractions recorded as part of
a reflection survey rather than a large number of LVL
surveys.
The same techniques can still be used for deeper
refractors, which require longer source-to-receiver offsets and a recording arrangement to give reversed coverage on the refractor. However, it may again be preferable to obtain the refraction data from the main reflection survey. As the depth increases, the offset distance
(XY value) becomes significant and must be accounted
for in the interpretation procedure, especially if structure is present at the level of the refractor. This requirement restricts the applicable methods to those based on
an emergent point analysis (see Figure 5-38b), such as
the GRM. Alternatively, one of the other interpretation
techniques listed below, which are more traditionally
used for continuous coverage data, can be used.
For refraction arrivals recorded as part of a reflection
survey, relevant interpretation methods include the
delay time approach (Section 5.6.6); the GRM (Section
5.6.5); the inversion, surface-consistent decomposition,
or tomographic technique (Section 5.6.9); and the wavefront approach or its ray-tracing derivative (Section
5.6.8). Under most circumstances, these surveys generate continuous coverage refraction data, normally with
a significant amount of recorded overlap between successive source locations, leading to a large surface
redundancy.
It is sometimes appropriate to use more than one
method to analyze the refraction data (e.g., Laidley and
Mills, 1986; Beattie and Wardell, 1987; Diggins et al.,
1988; Zanzi, 1990). For example, the high- and low-spatial-frequency components can be estimated indepen-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

230

Static Corrections for Seismic Reflection Surveys

dently. The intercept-time method (Section 5.6.1) either


produces suboptimum results or is impractical if there
are high-frequency (short-wavelength) time shifts present on the timedistance curves. Thus, a possible
approach is to evaluate and apply the high-frequency
variations first and then use the intercept-time method.
This approach is also appropriate for the wavefront
method (Section 5.6.8), even though the technique is
capable of operating from the surface, such that the shallow information is derived prior to an analysis of the
deeper data. Thus, the first pass could estimate the nearsurface anomalies and reduce the large number of data
points to an average set to produce one composite
timedistance curve for the whole line.
For the delay time method (Section 5.6.6) and generalized reciprocal method (GRM) (Section 5.6.5), the offset distance or XY value requires computation. More
than one value is often present in the data. A value of
zero or near zero is used for the near-surface layers and
a nonzero value for any deeper seated features, such as
topography on a deep refractor. Two separate analyses
are required to achieve this, and the difference is based
on the depth of an anomaly rather than on a spatial-frequency basis (described above). In the inversion methods (Section 5.6.9), for stability reasons it is preferable to
generate a low-frequency (long-wavelength) near-surface model, which is followed by the computation of the
high-frequency components with a surface-consistent
delay time analysis.
Diggins et al. (1988) proposed a multistage procedure
with the objective of minimizing long-wavelength
errors introduced in a decomposition approach from
possible errors in the refractor velocity (see also Archer
and Heathcote, 1985). An initial set of receiver delay
times are estimated using an interpretation technique
that does not require refraction velocity information,
such as the ABC, plus-minus, or GRM; these delay times
are also assumed to apply at collocated or closely located source locations. The refractor velocity is estimated
from these delay times and the original arrival times
using a rearrangement of equation (5.99) (Section
5.6.9.2). This velocity is then used to compute source
and receiver delay times on an iterative basis using
equation (5.100) and an equivalent equation for the
receiver delay times.
I described different types of near-surface topography in Section 2.4, including sand dunes, areas with
highly irregular weathered layers, youthful and mature
topography, permafrost, and mountain front topography. Two possible target depths for refraction surveys
are likely in many areasthe base of the weathered
layer and deeper refractors. For example, youthful and
mature topography often fall into this category. The various interpretation methods referred to earlier for refrac-

tions recorded as part of a reflection survey can be


applied to most of these areas. The main exception is
that a full interpretation in mountain front topography
is often impractical because of the presence of steep
dips. In addition, the refraction technique may be inappropriate in some permafrost areas. The thickness of the
permafrost and the velocity and thickness of the underlying sediments will determine whether first arrivals are
observed on the recording spread. Other factors may
need to be considered in some areas. For example,
where there is topography on the refractor, the offset
distance (XY value) can vary along the line due to significant elevation changes. This factor is more important
in an analysis of deeper refractors, which have a significant offset distance, or of deeply buried features in
mature topography.
In areas where the refractor dip is steep, many of the
refraction interpretation methods may no longer be
applicable. For example, the limit for the delay time
method is normally quoted at about 10 and for the
GRM at 20 (Palmer, 1980). Steeper dips and a variable
offset distance can be handled with the wavefront
method or one of its ray-tracing derivatives. However,
generation of the composite timedistance curve is likely to be difficult because steep and variable dips imply
potentially rapid changes in the crossover distance. In
such areas, the apparent velocities can range from
abnormally slow to fast and even negative values, and
hence it is not easy to identify the relevant segments
from the individual timedistance curves correctly. In
addition, good velocity control is required down to the
mapped refractor.
In mountain front topography, where the near surface is complex or where complex tectonics are
involved, refractor dip is likely to be steep and the
refractors may outcrop or subcrop along the line. This
implies that several refractors may need to be mapped
and that these may change rapidly along the line. These
complexities pose such major problems for the refraction technique that it is rarely used in these areas for a
full interpretation, although less complex portions of a
line are often analyzed on a piecemeal basis.
Information from a geologic survey or a shallow reflection survey can help define the parts of a line where a
refraction survey or refraction arrivals may give useful
information. This may be limited to defining the velocities in the outcrops. In the subsequent definition of the
near-surface model and computation of datum static
corrections, a large amount of interpretive judgment is
normally required. It may be appropriate to conduct a
sensitivity analysis on some data to determine the range
of datum static corrections that can be computed from
the available information. I discuss other aspects of
near-surface interpretation in Sections 3.4 and 3.8.7.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Baixas et al. (1985) classified refraction problems into
three types:
1. gently dipping markers, with homogeneous overburden;
2. multilayered overburden overlying irregular
refractors such that variable offset distances are
involved; and
3. areas of complex tectonics, with changing refractors along the line.
The first type is routinely handled by all refraction interpretation methods; I previously described some of the
problems associated with the other two categories.

5.7

VELOCITY CONTROL AND DEPTH


CONVERSION

The construction of a near-surface model using


refraction arrival or delay times requires the near-surface velocity (and with most interpretation methods, the
refractor velocity) to convert the times to depth. The
conversion of delay times to depth was defined by equations (5.64) and (5.65) for the single-layer case (Section
5.6.2). Additional equations were included in Section 5.3
and in specific interpretation methods described in
Section 5.6.
Velocity information can be obtained from an uphole
survey, the refraction data itself, or a reflection survey. I
describe in Section 5.7.1 various ways to acquire nearsurface velocities and in Section 5.7.2 refractor and subweathered layer velocities. The latter includes (in
Section 5.7.2.2) a few points on the conversion of refraction velocities to vertical velocities, which are required if
the near-surface model is to be used to compute datum
static corrections. I review the depth conversion stage in
Section 5.7.3, which includes a discussion of errors, both
in refractor depth and datum static corrections derived
from the near-surface model.
The physical quantities measured on uphole and
refraction surveystime, depth, and distanceare vulnerable to errors, which is true for any physical measurement. This can lead to significant errors, such as in
interval velocity estimates. These errors, which may be
random or more organized with a dc bias, impact the
various computational steps, such as delay times, velocities, and the final near-surface model. The complexity
of the earth is inherent in the total error, often such that
the data set is undersampled. This is certainly true for
the near-surface velocity in many areas, as it is generally uneconomic to obtain adequately sampled data. The
layering observed in uphole surveys along a line often
gives an indication of the spatial near-surface variations;

231

in some areas they can be easily correlated, while in others there is little or no apparent correlation.
Even when refracted arrivals recorded as part of the
reflection survey are used, in which refraction delay
times can be obtained with a spatial sample of the group
interval, a near-surface velocity field is still required to
convert these to depth. This is normally obtained with
much coarser sampling, necessitating a large amount of
interpretive judgment in the spatial interpolation to
minimize the error in the final near-surface model. In
these areas, the error in the near-surface or weathered
layer constitutes a major part of the total error.
There is often an ambiguity between the refractor
depth and near-surface velocity, often requiring the
interpreter to decide between a smooth refractor and an
irregular near-surface velocity profile, or the converse; I
discuss this in Section 5.7.3.1. Section 5.7.3.3 includes a
brief discussion of the impact of near-surface velocity
errors on the near-surface model and subsequent
derivation of datum static corrections. This shows that
some errors can be tolerated in the velocity estimates,
providing the generation of the model and the computation of the datum static corrections use the same velocity information. As a general rule, it is advisable to use a
relatively simple near-surface model unless there is
good evidence for a more complex one. This follows the
principle of Occams razor referred to in Section 3.8.7.

5.7.1

Near-Surface Velocities

Several methods are available for the computation of


velocity information from the surface down to the
refractor. Values are generally obtained at specific locations which are then used with other estimates to build
up a picture of the spatial velocity variation over the survey area (see Sections 3.4 and 3.8.7). I noted earlier that
the near-surface velocity may change rapidly along the
line such that it is inadequately sampled by the measurements made during the survey. One or more layers
of constant velocity may occur at each location, or there
may be a velocity increase with depth within each layer.
In sand dune topography, for example, the average
velocity from the surface to the refractor can vary with
the thickness of the sand as a result of compaction (see
Section 2.4.1). Section 5.7.3 discusses the impact of nearsurface velocity errors, both on the computation of the
weathering thickness from a delay time and on vertical
time through the near-surface layers in the computation
of datum static corrections.
Near-surface velocity information is generally
obtained from an uphole survey (Section 5.7.1.1), which
often gives the most reliable information, or from the
direct arrivals recorded on a refraction survey (Section
5.7.1.2). Velocity information can also be obtained from

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

232

Static Corrections for Seismic Reflection Surveys

a shallower refractor mapped during the survey


(Section 5.7.1.3), an analysis of the delay times along the
line (Section 5.7.1.4), or a shallow reflection survey
(Section 5.7.1.5). For example, Reynolds et al. (1990)
showed cross-sections on which a dense coverage of
refractor velocity information from several layers was
contoured.

5.7.1.1 Uphole Surveys


Uphole surveys are designed to evaluate the thicknesses and velocities of the near-surface layers in a borehole. I gave an extensive description of the technique,
including acquisition methods and interpretation, in
Chapter 4. I noted that the drilling process itself alters
the formations close to the borehole, so that all necessary
steps should be taken to ensure that the differences
between the interpreted velocities and those of the unaltered formations are minimized. As indicated in Section
4.4.3, however, this is not easily achievable in some situations. Potential errors in the estimated velocities are
likely to be greatest for near-surface layers, especially
when water or mud is used in the drilling process
through dry formations. However, good near-surface
velocity information is generally obtained from carefully operated and controlled uphole surveys and is often
better than that from a refraction survey, especially if
hidden layers are present.
The velocity information required is from the surface
down to the refractor, although in some cases, the refractor is not easy to identify on the uphole display. In these
situations, a possible starting point can be established by
converting the time information from the refraction survey at the uphole location to an approximate vertical
time using a cosine factor. These corrected times can then
be compared with the uphole survey times. This analysis
may involve an individual survey location or may
encompass a large number of surveys over the area.
The estimated velocities apply only at the uphole survey location, which may not be typical of the surrounding area; this error may be more significant than timing
or depth errors. I showed in Section 4.4 that the magnitude of interval velocity errors depended both on the
individual data values and on the thickness of the layer,
with the error increasing as the thickness of the layer
decreased.
A limited amount of near-surface velocity information can be obtained from the uphole times recorded on
a deep-hole dynamite survey (see Section 4.6). Adequate
QC checks must be in place to ensure that reliable
uphole times and charge depths are obtained. With only
one time estimate, the information obtained is the average velocity from the surface to the location of the dynamite charge. The most appropriate average near-surface

velocity is obtained where the charge is at or close to the


base of the weathered layer. In some cases, the location
of the charge with respect to the base of the weathered
layer can be obtained from the drillers report; this is
most likely if there is a distinct interface between the
weathered and subweathered layers. I noted in Section
4.6 that this information could be supplemented in some
areas by an analysis of the record quality and ground
roll. If a charge is known to be below the base of the
weathered layer, the average velocity obtained will generally be faster than the weathered layer itself. If, however, all the charges along a line are in the weathered
layer, the velocity information is less definitive,
although it can sometimes be used as an aid in establishing velocity trends between deep uphole survey
locations.

5.7.1.2 Direct Arrivals


The short offsets on a weathering or low-velocity
layer survey often record the direct arrival through the
near-surface layer as a first arrival (see Section 5.4.1).
However, in the case of refracted arrivals recorded as
part of a reflection survey, the minimum offset used may
be such that the direct arrivals are not recorded as first
arrivals. In this situation, an assumption can be made
that the minimum offset corresponds to the crossover
distance between the direct and refracted arrivals; this
can then be used to obtain an estimate of the near-surface velocity, although any derived velocities must be
analyzed carefully. If this assumption is wrong, that is,
the minimum offset is greater than the crossover distance, this approach leads to an overestimate of the nearsurface velocity.
The origin of the timedistance curve can be used as
a data point in the analysis of the direct arrival, providing all necessary adjustments are made to the first-break
times to convert them to absolute times. This must allow
for any time break correction or system delay (see
Section 5.5.7). In areas with large elevation changes
between the source and receiver, it is necessary to correct for the (assumed) sloping travel path. This is on the
basis that the group interval is surveyed as a horizontal
distance and is not a distance measured along the surface. In some areas, it will be found that a velocity
increase with depth fits the plotted points better than
that of a constant velocity (Section 5.3.5).
The velocity information obtained from direct
arrivals is considered to apply between the source location and the farthest offset used in the analysis; it is an
average value or function over a range of surface locations. In areas with rapid lateral changes in near-surface
velocity, this may be theoretically inappropriate, but it is
generally preferable to use an average value rather than

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)
SP 1
50

5.7.1.4 Delay Time Analysis


A crossplot analysis of delay time (defined in Section
5.6.2) and surface elevation can be used to estimate an

Elevation (m)

30

Surface

Weathered layer
Model 1
Model 2
(b)
50

5.7.1.3 Shallower Refractor Information

25

0
(c)
50

50

V = 800 m/s

Model 2

Elevation (m)

Model 1

Elevation (m)

If refracted arrivals from a refractor shallower than


the one being mapped are available, they can be used to
give additional velocity information and to supplement
direct arrival or uphole survey velocities. However, as
noted in Section 5.7.1.2, hidden layers may be present,
with the result that an approximate velocity function
down to the refractor is obtained, as it does not incorporate all the layers. The refractor velocity is estimated
with one of the refraction interpretation techniques
described in Section 5.6. These include the intercepttime, plus-minus, generalized reciprocal method
(GRM), delay time, wavefront, inversion, and tomographic approaches. I include further comments on
these velocity approaches in Section 5.7.2.1, along with
some reminders on factors affecting the accuracy of the
interpretation.
The refraction velocity normally represents the velocity parallel to the bedding planes. Due to the effects of
anisotropy, this is faster than the velocity perpendicular
to the bedding planes for most formations. The latter
velocity, or more correctly, the vertical velocity, is
required for subsequent computation of the layer thickness down to the target refractor. I discussed this factor
in Section 2.6.5 on anisotropy (see also Section 5.7.2.2),
which showed that the velocity perpendicular to the
bedding plane was typically about 8590% of that parallel to the bedding plane.

20

10

Delay time (ms)

a precise one that is good only at one location. When an


analysis of diving or turning waves in the near-surface
layer is used in an inversion or tomographic technique
to estimate the near-surface velocity (Sections 5.6.9.1
and 5.6.9.3), some lateral velocity variations may be
obtained.
When the direct arrival is used to convert refraction
times to depth, it is possible that this may misrepresent
the velocity from the surface to the refractor. This can
occur if additional layers are present between the surface layer and the refractor. In some cases, additional
refractors can be mapped in this interval (see Section
5.7.1.3), so that their velocities can be incorporated in the
near-surface velocity model. Even with such additional
control, hidden layers such as thin layers or velocity
inversions may be present, leading to an error in the
average velocity from surface to the refractor. Section
5.7.3 discusses the impact of possible errors from incorrect near-surface velocities on both the depth model and
datum static corrections derived from this model.

233

V = 1140 m/s
0

0
50

0
Delay time (ms)

50

0
Delay time (ms)

Fig. 5-82. Near-surface model to illustrate delay time


analysis for the computation of the near-surface velocity:
(a) two near-surface models with a common surface elevation but different weathering thicknesses; (b) delay
time profiles for the two near-surface models; (c) crossplots of elevation against delay time for the two near-surface models.

average near-surface velocity in some areas. The


requirements are an elevation profile with an appreciable short-wavelength (high-spatial-frequency) component and a refractor depth profile that is known (or can
be assumed) to be smooth and preferably planar. Delay
times are estimated by techniques such as ABC, plusminus, generalized reciprocal method (GRM), and delay
time methods. The technique is illustrated with the
model data shown in Figure 5-82. In model 1 (Figure
5-82a), the base of the weathered layer is flat, resulting in
the crossplot shown in Figure 5-82c. The slope of this

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

234

Static Corrections for Seismic Reflection Surveys

crossplot gives the correct velocity of 800 m/s. Before


this velocity can be used to convert times to depth, it
must be modified by a cosine term to generate the nearsurface velocity. This can be defined as
V = Va cos c ,

(5.102)

where V is the near-surface velocity, Va is the velocity


measured by the delay time analysis technique, and c is
the critical angle between the near-surface layer and the
refractor.
Figure 5-82a shows that the base of the weathered
layer for model 2 forms a gentle anticline, with a relief of
just over 10 m. The crossplot for these data (Figure
5-82c) indicates a velocity of 1140 m/s, about 40% above
the correct velocity. In this case, the delay times no
longer relate simply to the surface elevation because the
refractor is not flat. The weathered layer thickness varies
by about 20 m along the profile, about 75% of the surface
elevation relief of 26 m. If the derived velocity of 1140
m/s is used to convert the delay times (Figure 5-82b) to
depth, the base of the weathered layer elevation varies
from 4 to 5 m instead of the correct value of 0 m. This
example demonstrates that the ideal situation for this
technique is where the surface has some relief but the
refractor is flat. In addition, erroneous velocities are
obtained when changes in surface elevation do not represent changes in weathered layer thickness.
Figure 5-83 shows three delay time analyses using the
data in Figure 5-60 from an area characterized by rapid
changes in near-surface velocity. The three analysis locations correspond to the three hills shown on the surface
profile in Figure 5-83a. The crossplots in Figure 5-83b are
of delay time against the elevation difference between
the surface and regional elevation, shown as a linear
trend. These should give useful velocities, providing the
base of the weathered layer has a similar slope to the
regional elevation profile. The results from the first two
locations are broadly in agreement with the uphole survey results shown in Figures 5-60b and 5-83a, but at the
third location (502516), the velocities are significantly
different. If these velocities are applied to the delay
times, the base of the weathered layer is shown in Figure
5-83d to dip at about the same angle as the regional elevation profile.
Another possible approach for the analysis of these
data is shown in Figure 5-83c, where the crossplots are
of delay time against surface elevation. This produces
bimodal distributions for the first two analysis locations,
with the first six locations in each analysis producing the
higher of the two velocities, but it gives a small scatter of
points for the third location. The resulting base of the
weathered layer shown in Figure 5-83d contains several
nearly horizontal segments resulting in two large discontinuities, each about 30 m, where the bimodal distri-

butions split at locations 475 and 485. This indicates that


the velocity changes within these analyses are unlikely
to be correct and that the velocities are different from
those in the uphole surveys.
The base of the weathering profile in Figure 5-60b indicated an almost constant slope over the first two analysis
locations and almost flat over the third. This supports the
above observations that the crossplots are better if the elevation of the surface above the regional is used at the first
two locations and the surface elevation used for the third
location.
This technique is appropriate in some areas, but it can
lead to erroneous velocity estimates and, as such, must be
used with caution. This means that a thorough QC procedure must be adopted, which should include a review
of the crossplots and a critical examination of the base of
the weathered layer elevation profile. Different assumptions can be made about the overall shape of the base of
the weathered layer and appropriate regional elevation
profiles used, so that the crossplots relate as closely as
possible to changes in weathered layer thickness versus
delay time. It is the variations in the weathered layer
thickness that do not correspond to elevation changes
which distort the results.

5.7.1.5 Reflection Surveys


Reflections from the base of or within the weathered
layer are observed on some reflection surveys. These
reflections are not necessarily from the same interface
mapped by the refraction survey. I noted in Section 5.1
that reflections require an acoustic impedance contrast,
whereas refractions occur with a velocity contrast (see
Table 5-1). In addition, a relatively thick layer is required
for refracted energy to be propagated successfully. The
observations can be from a separate shallow high-resolution survey or from the main reflection survey and can
be used to generate the near-surface model (see Sections
3.8.5 and 3.8.7).
The two-way reflection time data can be used to
derive velocity information down to the reflectors if
multifold data are recorded at the shallow target.
However, this requires a good set of datum static corrections, especially for the short-wavelength component, so that an acceptable common-midpoint (CMP)
stack response is obtained. This may require an iterative
approach to establish the best velocity for the datum static corrections, especially in areas where there are rapid
lateral near-surface changes. Alternatively, several different near-surface velocities can be tried, and the one
producing the best CMP stack response used.
If the base of the weathered layer is observed, its
stacking velocity can be used to give an estimate of the
average velocity from the surface to the reflector.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

235

(a)
120

Elevation (m)

580

Surface

80

695
450
830
40

Regional elevation

635
1000

0
Location

470

490

480

500

510

520

(b)
40

880 m/s

20

40

Locations 480 - 496


Elev (m)

Locations 469 - 480


Elev (m)

Elev (m)

40

860 m/s
20

Locations 502 - 516

20
370 m/s

0
0

Delay time (ms)

50

0
0

Delay time (ms)

50

Delay time (ms)

50

(c)
80

100

Locations 469 - 480

120

Locations 480 - 496

Locations 502 - 516

670 m/s
1450 m/s

40

20

1190 m/s
60

100

40
0

(d)

80

Elevation (m)

Elevation (m)

Elevation (m)

340 m/s
60

Delay time (ms)

50

890 m/s
80

60
0

Delay time (ms)

50

Delay time (ms)

50

120

Surface

Elevation (m)

80

Base of weathered layer


Velocities from (b)

40

Velocities from (c)

-40
Location

470

480

490

500

510

520

Fig. 5-83. Computation of near-surface velocity information using a delay time analysis technique on the data shown in
Figure 5-60: (a) surface elevation profile (from Figure 5-60b), regional elevation profile, and uphole survey velocities; (b)
crossplots of elevation difference between surface and the regional elevation against delay time; (c) crossplots of surface elevation against delay time; (d) base of weathered layer profiles using velocities from (b) and (c).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

236

Static Corrections for Seismic Reflection Surveys

However, in the presence of dip, the velocities should be


adjusted by the cosine of the dip, unless dip moveout
(DMO) is applied to the data prior to the analysis. In
complex tectonic areas, it may be necessary to use a
more rigourous technique, such as a ray-tracing
approach, to estimate the interval velocities.
If several reflectors are mapped, the interval velocity
in the nth layer (VIn) can be computed using Dixs formula (Dix, 1955):
V 2 t Vn2 1tn 1
VI n = n n

tn tn 1

1/ 2

(5.103)

where Vn and Vn1 are the rms velocities for the reflectors above and below the layer and tn and tn1 are their
reflection arrival times. The stacking velocity estimated
by a velocity analysis in most situations is almost equal
to the rms velocity.
The derived interval velocities will have larger error
ranges than those of the individual stacking velocity
estimates; the magnitude of these should be taken into
account in any subsequent analysis. Schneider (1971)
stated that in the presence of ambient noise, the interval
velocity to rms velocity error was about 1.4 times the
ratio of bed depth to bed thickness. This error can be
reduced if spatial averaging is applied to the estimates,
but this is not appropriate in areas where the near-surface velocity changes rapidly along the line.

5.7.2

Refractor and Subweathered


Layer Velocities

The conversion of refraction arrival times to depth


requires an estimate of the refractor velocity for most of
the interpretation techniques. This is routinely obtained
from an analysis of the refraction arrival times (see
Sections 5.6 and 5.7.2.1). The subweathered layer velocity is required for any subsequent computation of
datum static corrections from the near-surface model.
I showed in Section 3.1 that the computation of
datum static corrections could be divided into two basic
steps: removal of the weathered layer, followed by an
adjustment from the base of the weathered layer up (or
down) to the reference datum. The velocity used for the
latter correction is usually called the replacement velocity,
or sometimes the datum velocity, elevation velocity, or subweathering velocity. The replacement velocity generally
changes slowly or is constant along the line (see Section
3.3). It can be estimated from refraction or uphole surveys (Sections 5.7.2.1 and 5.7.2.3). Information can also
be obtained from a shallow reflection survey (Section
5.7.2.4). The velocities are then smoothed and/or interpolated to generate the space invariant or space variant

replacement velocity profile. If refraction velocities are


used, they must be converted to a vertical velocity using
information from uphole surveys or an appropriate correction factor (Section 5.7.2.2). Section 5.7.3 covers the
impact of refraction velocity errors in the computation
of weathering thickness and datum static corrections.

5.7.2.1 Estimates from the Refraction


Interpretation Method
The refractor velocity is estimated from the corrected
first-arrival times using almost any of the refraction
interpretation techniques described in Section 5.6. In
weathering or low-velocity layer surveys, this includes
the intercept-time, plus-minus, and generalized reciprocal methods (GRM). The intercept-time method (Section
5.6.1) assumes that the trace-to-trace changes in the
weathered layer are small, that is, minimal short-wavelength (high-frequency) time shifts along the recording
spread. The updip and downdip apparent velocities are
estimated from the slope of the timedistance curves
from the two recording directions. The true velocity is
computed after estimating the critical angle and refractor dip using equations (5.18) and (5.21) (Section 5.3.3),
or for small angles, the two velocities can simply be
averaged. If the refractor velocity is evaluated using the
plus-minus method of Hagedoorn (Section 5.6.4) or the
GRM (Section 5.6.5), the effect of near-surface related
static shifts is eliminated. The velocity is obtained directly from a plot of the minus times against distance or
from the velocity analysis functions.
For refraction arrivals recorded as part of a reflection
survey, interpretation techniques such as the delay time
approach (Section 5.6.6), the GRM (Section 5.6.5), an
inversion, decomposition, or tomographic technique
(Section 5.6.9), or a wavefront approach or its ray-tracing derivative (Section 5.6.8) are generally used. In most
cases, continuous coverage data is obtained that allows
for the computation of a refractor velocity profile along
the line. I noted in the description of the different techniques that it was sometimes necessary to identify and
remove time shifts associated with near-surface anomalies before the deeper data could be analyzed. In the
case of the delay time approach, for example, surfacerelated time shifts show up on the intercept-time curves
in different positions for the two directions of recording
if a nonzero-offset distance is used in the analysis. Thus,
if near-surface anomalies are present and not removed,
it is necessary to smooth the intercept-time curves to
obtain the refractor information and the underlying
velocity trend.
Care must be taken to ensure that all potential errors
are minimized. This includes errors associated with
picking the pulse (see Section 5.4.3), the effects of noise,

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


amplitude decay (including shingling), arrays, and
intra-array static corrections. I showed in Section 5.5.1
that time picks were influenced by the gain of the display. In addition, if a phase of the arrival later than the
first break is picked, the common assumption of a constant time correction can introduce errors if there are significant near-surface changes along the line. Section
5.5.7 documented the corrections that should be applied
to arrival times, such as time break corrections, recording instrument delay, uphole time corrections, source
and receiver array corrections, and offset corrections.

5.7.2.2 Conversion to Vertical Velocity


The velocity estimated by a refraction survey normally represents a horizontal velocity, or more correctly,
a velocity parallel to the bedding planes. For computation of datum static corrections, a vertical velocity is
required. I showed in Section 2.6.5 (on anisotropy) that
the velocity perpendicular to the bedding planes was
typically about 8590% of that parallel to bedding,
although the range of values was about 70100%. This
factor can be evaluated at various locations along a line
if both uphole and refractor velocities are available.
However, problems can arise in correlating the results
from these two different surveys because of discrepancies in the depth information. Discrepancies can be
caused by factors such as observations from different
layers (sometimes because a relatively thick layer is
required for a refractor to be observed, especially at long
offsets), hidden layer effects, and various systematic
timing errors. If this comparative information is not
available, the normal approach is to apply a correction
factor of 90% to the refraction velocities to obtain the
equivalent vertical velocity.

5.7.2.3 Uphole Surveys


I stated in Chapter 4 that, whenever practical,
upholes should be drilled into the consolidated layer
below the weathered zone. Providing there are sufficient
time-depth values recorded in this layer, a good estimate of the subweathering velocity or velocities can be
obtained. In some cases, the top of this layer corresponds to the refractor mapped by a refraction survey.
The correlation is not always obvious, and account must
be taken of the likely difference in velocities. This is
because the uphole survey measures a vertical velocity,
and the refraction velocity is normally the component
parallel to the bedding planes (see Sections 2.6.5 and
5.7.2.2). I noted in Section 5.7.1.1 that a starting point can
be estimated by converting the refraction arrival time at
the uphole location to a vertical time so that it can be
compared directly with the uphole survey time.

237

In most areas, the velocity variation away from the


uphole survey location is less in the consolidated material than in the near-surface layer. Thus, the point made
in Section 5.7.1.1 about the measured velocities possibly
being atypical of the surrounding area is less likely to be
true for subweathered layer velocities. The interval
velocity error depends both on the time and depth
errors of the individual data values and on the thickness
of the layer (see Sections 4.4 and 5.7.1.1).

5.7.2.4 Reflection Surveys


The subweathered layer velocity can be estimated
from shallow seismic reflection data in some areas. This
requires that a reflection close to the base of the weathered layer and a slightly deeper reflector can be mapped
and their stacking velocities estimated. I noted in
Section 5.7.1.5 that good short-wavelength datum static
corrections are required for velocity analyses. This may
necessitate an iterative or trial-and-error approach prior
to the subweathered layer velocity evaluation. The various techniques used to acquire the data, which can be a
separate shallow high-resolution survey or the main
reflection survey, are described in Section 3.8.5.
The interval velocity is derived using Dixs formula,
as given by equation (5.103) in Section 5.7.1.5, where the
assumption normally made is that the stacking velocity
estimated by a velocity analysis is almost equal to the
rms velocity. Various comments on corrections that may
be required to account for dip are given in Section 5.7.1.5.
I also referred in Section 5.7.1.5 to the magnitude of the
interval velocity errors and indicated that these were significantly higher than those of the rms velocity errors.
Spatial averaging to reduce the interval velocity error is
generally more appropriate at this level than for nearsurface velocities because they are more likely to change
slowly along the line. However, this type of smoothing is
inappropriate in areas with complex geology.

5.7.3

Depth Conversion and Impact of


Errors

The refraction time and velocity information is used


to compute the depth of the near-surface layer(s) along
the line. The conversion of delay times to depth is given
by equations (5.64) and (5.65) in Section 5.6.2. For some
interpretation techniques, the depths are plotted directly below the observation point, while in others they are
plotted from the surface at a location vertically above
the emergent point from the refractor. The latter represents a migration procedure and incorporates the XY
value of the generalized reciprocal method (Section
5.6.5) and the offset distance on the delay time method

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

238

Static Corrections for Seismic Reflection Surveys


(a)

Elevation (m)

Location
100

485

490

495

485

490

495

50

0
(b)

Elevation (m)

Location
100

50

Fig. 5-84. Refractor depth profile of short section of line shown in Figure 5-60: (a) construction using arcs, with a radius
equal to the thickness, centered on surface locations; (b) conventional profile where depth (thickness) information
assumed to relate to point vertically beneath the surface location.

(Section 5.6.6). These values are often zero or close to


zero when dealing with near-surface refractors.
In many cases, the computed depths are estimated as
thicknesses perpendicular to the refractor, making it
necessary to adjust for the local dip of the refractor prior
to computation of the depth profile. This can be done
mathematically, as shown in Section 5.3.3 for a two-layer
dipping interface, or by a simple graphical approach.
For the two-layer case, arcs with radii equal to the computed depth and centered on the surface location are
constructed, and the tangent to these represents the
refractor. An example is shown in Figure 5-84, which
uses part of the near-surface model displayed in Figure
5-60. In the case of the wavefront method (Section 5.6.8),
however, the depth profile is constructed directly from
the observed traveltimes and near-surface velocity
information. After depth conversion, the elevation profile of the refractor is integrated with other data, such as
deep upholes and information from intersecting lines
(Section 3.8.7).
I discuss the ambiguity between refractor depth and
near-surface velocity in Section 5.7.3.1; interpretive
judgment is normally required to resolve this ambiguity. Section 5.7.3.2 examines the various errors that can be
involved, which includes many of the points made earlier in this chapter. Section 5.7.3.3 discusses the impact of
such errors, especially those related to the near-surface
and refractor velocities, on subsequently derived datum
static corrections.

5.7.3.1 Refractor Depth and Near-Surface


Velocity Ambiguity
An accurate refractor depth profile requires both
accurate refraction times and velocity information. I
noted earlier that the near-surface velocity is normally
estimated at discrete locations along a line, which means
that the data may be undersampled in areas where
rapid variations occur along the line. In subsequent
interpolations of these values (see Section 3.4), the initial
assumption is generally that the variation between the
control points is fairly smooth. The resulting refractor
depth profile may contain high-spatial-frequency
(short-wavelength) variations, which may be correct.
However, they may represent errors due to observational errors or inadequate sampling of the near-surface
velocities.
Such variations can be ignored if one assumes that
the information is correct or that the derived datum static corrections will be accurate enough for the survey.
Alternatively, they can be smoothed to produce a different near-surface model. In this case, the near-surface
velocity profile must be adjusted so that it forms part of
a consistent data set. That is, the surface elevation,
refractor elevation, and the near-surface velocity must
yield the observed refraction time information, such as a
delay time or intercept time.
The decision as to which near-surface model to
usethat of a smooth near-surface velocity profile and

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


an irregular refractor, a smooth depth profile and an
irregular velocity profile, or an intermediate near-surface modelis an interpretive one. It should be based
on a geologic understanding of the area. Baixas et al.
(1985) and Daly and Diggins (1988), for example, suggested that high-frequency variations in the base of the
weathered layer could be indicative of an incorrect
near-surface velocity. Daly and Diggins (1988) referred
to the approach as a residual refraction statics procedure and that the overall objective was a more
informed decision about how the refraction traveltime
variability should be distributed between the near-surface velocity and the depth of the weathering layer
(Daly and Diggins, 1988, 574).
Archer and Heathcote (1985) described two sets of
ambiguities: that between refractor structure and shallow velocity anomalies, and between refractor velocity
variations and shallow velocity anomalies. They
showed that the ambiguities occurred at a range of
wavelengths related to the critical distance, which
meant that they could be minimized by simultaneously
mapping several refractors with different critical distances. In his analysis of the accuracy of several refraction interpretation methods, Zanzi (1990) defined the
latter ambiguity in terms of the weathering thickness
and refractor velocity. This was on the basis that the
near-surface or weathering velocity must be established
from an external source of information and is therefore
a known quantity for the refraction interpretation.
The possible near-surface models described above
can be produced from the same refraction times; however, these models result in different datum static corrections because the refractor depth changes. This is
illustrated with data from Michigan where a deep
weathered layer is present. Two possible near-surface
models are shown in Figure 5-85b for the delay time
profile in Figure 5-85a. In model A, the refractor elevation varies from about 55 to 115 m, with a near-surface
velocity of 1500 m/s. In model B, the flat refractor is at
an elevation of 90 m and the near-surface velocity varies
from about 1500 to 2000 m/s. The resulting datum static corrections, to a datum at 180 m, are shown in Figure
5-85b. These show variations between the two near-surface models of about 10 ms (5 to +5 ms) along the line.
The preferred model, based on a knowledge of the area,
is model B with the smooth refractor and short wavelength variations in the near-surface velocity. This decision is supported by a comparison of the stacked data
shown in Figure 5-85c, in which the smoother reflection
times with the datum static corrections from model B
are judged to be more representative of the subsurface
geology in the area.
Figure 5-86 shows two possible near-surface models
for the data shown in Figure 5-60, which is from an area

239

characterized by rapid near-surface velocity changes


along the line. The smoother refractor in Figure 5-86a
corresponds to the erratic near-surface velocity profile in
Figure 5-86b. Some of this velocity variation is to be
expected in areas where the weathered layer is thin
because a 1 m change in thickness results in a large
change in near-surface velocity.

5.7.3.2 Observational Errors and


Uncertainties in Refraction
Interpretation
The correctness of the near-surface model derived
from refraction data is dependent on many factors,
including the accuracy of the field data measurements
(time picks and geometry of the recording), the correct
identification of refractor segments on a timedistance
curve, and whether there is adequate velocity control to
obtain accurate depth profiles. In addition, the complexity of the subsurface must be within the capability of the
refraction method and the technique used for the interpretation. Errors have been discussed by Hagedoorn
(1955), Domzalski (1956), Northwood (1967), Sjgren
(1979), Dampney and Whiteley (1980), and Laidley and
Mills (1986), among others.
The need for accurate time picks and their associated
corrections is of paramount importance, as noted in
Sections 5.4 and 5.5. The accuracy of the time picks was
shown to depend on many factors, including the signalto-noise ratio, interference from other arrivals, shingling, arrays, and intra-array static corrections. In addition, if a later phase of the first arrival is picked instead
of the first break, further errors can be introduced if
inadequate compensation is made for any spatial and
offset time variations between these two times. The corrections involved include the time break correction,
recording instrument delay, uphole time, source and
receiver array corrections, and offset corrections.
Various QC procedures were discussed in Section 5.5,
including validation checks to assist in the identification
of mispicks and displays of the picked times and refraction data. Problems identified can be due to picking
errors or their associated corrections or due to field
geometry errors, in which the actual recording arrangement and survey information differ from those specified
in the refraction analysis routine.
Refractor identification is a trivial task in many areas,
but as the refractor complexity increases, it becomes difficult to ascertain whether a change in apparent velocity
is due to the presence of a new refractor or represents a
change of dip. In survey areas with faults or steep and
variable dip, the correct pairing of the refraction arrivals
from the two directions of recording is often not obvious, leading to further possible errors.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

240

Static Corrections for Seismic Reflection Surveys

( )
Delay time (ms)

(a)

50

100

150

(b)
Model A

Model B
Velocity (m/s)

1500

3000

1500

3000

4500

300

300
Elevation (m)

4500

200

200

100

100

-100

-100

Static correction (ms)

Static correction (ms)

Elevation (m)

Velocity (m/s)

-80

-60

(c)

-80

-60

Time (s)

Model A

Model A

Model B

Model B

0.0

0.0

0.5

0.5

1.0

1.0

Fig. 5-85. Data from Michigan to illustrate refractor depth and near-surface velocity ambiguity: (a) delay time profile;
(b) two possible near-surface models, A and B, which match the delay times and their associated datum static corrections; (c) common-midpoint stacked data with datum static corrections derived from models A and B (after Daly and
Diggins, 1988).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


(a)

Elevation (m)

Location
120

470

480

490

80

40

0
(b)

Velocity (m/s)

1500

500

510

520

241

Fig. 5-86. Near-surface


models derived from
the refraction data
shown in Figure 5-60:
(a) near-surface profile
with two possible
interpretations of the
base of the weathered
layer; irregular profile
corresponds to one
shown in Figure 5-60b;
(b) near-surface velocity profiles; linear segment profile corresponds with values
annotated in Figure
5-60b.

1000

500

In defining traveltimes through near-surface layers,


the refraction method works well in many areas, especially when the refractor velocity and depth profile are
slowly varying. This assumes that sufficient and accurate refraction arrival information is available, preferably with reversed coverage data. As the complexity of
the near surface increases, several of the interpretation
methods cease to be appropriate. For example, some are
limited to dips of about 10 (see Section 5.6). In areas
where the near surface is complex, including those with
steep dips, the refraction method is likely to become
impractical. Various comments on the choice of the
refraction interpretation method were included in
Section 5.6.13.
There are, however, many near-surface models that
correspond to a set of observed refraction arrival times.
For example, Hagedoorn (1955) and Ackermann et al.
(1986) showed timedistance curves that could be
explained by several near-surface models. Saleh (1994)
described some of the potential pitfalls and ambiguities
when refraction data are analyzed over a specific offset
range. Another aspect of the nonuniqueness of any
refraction interpretation is the inability of the refraction
method to map velocity inversions or thin layers (hidden layers). This leads to possible depth errors unless
velocity information down to the refractor is available
from another source, such as from an uphole survey.
The various sources of velocity information needed to
convert the refraction time information to depth were
described in Sections 5.7.1 and 5.7.2. The ambiguity
between near-surface velocity and refractor depth dis-

cussed in Section 5.7.3.1 again stressed the nonuniqueness of a refraction interpretation.


I have briefly discussed the topic of anisotropy, mainly in the conversion from horizontal to vertical velocities
or, more correctly, from velocities parallel versus perpendicular to bedding. Section 5.7.2.2 discussed this factor, which is required to convert the refractor velocity to
a replacement velocity. In addition, the potential problems associated with changes in velocity with azimuth
in crooked-line surveys were described in Section 5.6.10.
Similar problems may occur in 3-D surveys (Section
5.6.11).

5.7.3.3 Errors in Datum Static Corrections


Derived from Refraction Data
The accuracy requirements of datum static corrections depends both on the survey objectives and on the
capabilities of any subsequent residual statics correction
method. I show in Chapter 7 that these capabilities normally depend on the size of the errors and the signal-tonoise ratio of the reflection data. However, I stated in
Section 3.8 that residual static corrections are best at trying to solve small residual time corrections, so their ability to fix problems must not be an excuse for a substandard set of datum static corrections. The accuracy
requirements should also be divided into short- and
long-wavelength components, with the crossover normally considered to be between one-half and one
spreadlength. This is important because residual static
corrections are more likely to be successful for the short-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

242

Static Corrections for Seismic Reflection Surveys


Table 5-13. Datum Static Corrections Derived from Refraction Delay Time of 83 ms;
Error Analysis Showing Impact of Varying Near-Surface and Refractor Velocities.

NearSurface
Velocity a

Total
Elevation
Static
Correction
Correction
(datum 100 m deep)
(ms)
(ms)

Total
Elevation
Static
Correction
Correction
(surface datum)
(ms)
(ms)

Refractor
Velocity b

Weathering
Thickness

Weathering
Correction

(m/s)

(m)

(ms)

Variable Near-Surface Velocity


500
1800
43
600
1800
53
700
1800
63
800
1800
74
900
1800
86
1000
1800
100
1100
1800
115
1200
1800
134
1300
1800
156

86
88
90
93
96
100
105
111
120

35
29
23
16
9
0
9
21
35

121
117
113
109
105
100
96
90
85

27
33
39
46
53
62
71
83
96

59
55
51
47
43
38
34
28
24

Variable Refractor Velocity


1000
1620
1000
1980
1000
2160
1000
2340

105
96
94
92

3
2
3
4

102
98
97
96

72
54
48
44

33
42
46
48

(m/s)

105
96
94
92

a Correct near-surface velocity is 1000 m/s.


b Correct refractor velocity is 1800 m/s.

wavelength (high-spatial-frequency) components, and


it is the long-wavelength (low-frequency) components
that affect the structural times.
It is thus very important that datum static corrections
are as accurate as possible for the long-wavelength components. The components greater than two or three
spreadlengths can be treated as velocity anomalies so
that they are accounted for during the depth conversion
process. This assumes that the signal-to-noise ratio is
sufficient for adequate estimates of the spatial velocity
variation to be analyzed. It is generally preferable to
place the emphasis on obtaining a good set of datum
static corrections so that the depth conversion procedure
is limited to correcting small residual anomalies.
When datum static corrections are computed from a
near-surface model derived from refraction data, the
possible errors need to be examined. The near-surface
velocity and refractor depth ambiguity, as well as the
refractor velocity and near-surface velocity ambiguity
(described in Section 5.7.3.1), must be considered.
Correct datum static corrections can be produced only
when the refractor depth and near-surface velocity are
correct. This is often hard to achieve, especially in areas
where there are rapid changes in the near-surface velocity; this normally implies that the velocity field is undersampled, often due to insufficient uphole surveys.
Swan and Booker (1984) suggested that datum static
corrections are surprisingly resilient to near-surface
velocity changes. However, errors are still present, and

it is the survey objectives that dictate the level of error


that is acceptable. The example in Section 3.8.5.3 (Figure
3-29 and Table 3-6) demonstrated that datum static corrections do depend on the near-surface velocity. This
example was used to show how a shallow reflection
profile could be converted to a set of datum static corrections. The elevation correction component depends
on the near-surface velocity because this influences the
depth to the base of the weathered layer. A similar effect
was observed in the example from Michigan shown in
Figure 5-85 (Section 5.7.3.1).
To put into perspective the magnitude of errors in
datum static corrections, Tables 5-13 and 5-14 indicate
errors in depth and derived datum static corrections for
a simple two-layer model. These are included to illustrate a few points, but the results must not be used as
definitive for other near-surface models. In addition, the
values are analyzed at a specific location, and it is generally necessary to consider errors as a function of wavelength (e.g., Archer and Heathcote, 1985).
In the top part of each table, the near-surface velocity
is in error, and this velocity is used to compute both the
thickness and the datum static corrections; in the lower
part, the refractor velocity is varied. In each case, the correct near-surface velocity is 1000 m/s with a weathering
thickness of 100 m. In Table 5-13, the correct refractor
velocity is 1800 m/s and in Table 5-14 it is 3000 m/s; the
refraction delay times are 83 and 94 ms, respectively.
Datum static corrections are derived to a datum at 100 m

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys

243

Table 5-14. Datum Static Corrections Derived from Refraction Delay Time of 94 ms;
Error Analysis Showing Impact of Varying Near-Surface and Refractor Velocities.
NearSurface
Velocity a
(m/s)

Refractor
Velocity b

Weathering
Thickness

Weathering
Correction

(m/s)

(m)

(ms)

Total
Elevation
Static
Correction
Correction
(datum 100 m deep)
(ms)
(ms)

Total
Elevation
Static
Correction
Correction
(surface datum)
(ms)
(ms)

Variable Near-Surface Velocity


500
3000
48
600
3000
58
700
3000
68
800
3000
78
900
3000
89
1000
3000
100
1100
3000
111
1200
3000
123
1300
3000
136
1400
3000
149
1500
3000
163

95
96
97
98
99
100
101
103
104
106
109

19
16
12
8
4
0
4
9
13
18
23

114
112
109
106
103
100
97
94
91
88
86

18
21
25
29
33
37
41
46
50
55
60

77
75
72
69
66
63
60
57
54
51
49

Variable Refractor Velocity


1000
2400
1000
2700
1000
3300
1000
3600
1000
3900

104
102
99
98
98

2
1
0
1
1

102
101
99
99
99

48
42
33
30
28

54
60
66
68
70

104
102
99
98
98

a Correct near-surface velocity is 1000 m/s.


b Correct refractor velocity is 3000 m/s.

below the surface and to one at the surface. The replacement velocity used to compute the elevation correction
is assumed to be 90% of the refractor velocity.
The tabulated information in Tables 5-13 and 5-14
shows that a near-surface velocity error of 30% translates into a weathering thickness error of over 30%
(3256 m). However, the error in the total datum static
correction to a datum 100 m below the surface is less
than half this amount (915%), with an absolute error of
915 ms. With datum at the surface, the time error is
unchanged because varying the datum elevation with
the correct replacement velocity is equivalent to a constant time shift of the values. The percentage errors do
increase, however, especially for the values in Table 5-13,
which shows errors of about 35%. These values indicate
that on a pro rata basis, a 30% near-surface velocity error
and a weathering thickness of 30 m result in a datum
static error of 34 ms.
The weathering correction error is dependent on the
critical angle between the near-surface layer and the
refractor. As the velocity contrast increases, the critical
angle and the weathering correction error decrease; this
can be seen by comparing data from Tables 5-13 and 514, which have critical angles of 33.8 and 19.5, respectively. However, the elevation correction, which adjusts

the data from the base of the weathered layer to the reference datum, must still be considered in the analysis.
For a given refraction delay time, the weathering
thickness computation is relatively insensitive to errors
in the refractor velocity. For example, an increase in the
refractor velocity of 20% (from 1800 to 2160 m/s), using
the near-surface velocity of 1000 m/s, is shown in Table
5-13 to result in a decrease in the weathering thickness
of about 6%. For the higher-velocity model in Table 5-14,
a 20% increase in refractor velocity (from 3000 to
3600 m/s) results in a decrease in the weathering thickness of about 2%. In this situation, the total datum static
correction is dependent on the datum elevation because
the replacement velocity is in error. With datum at 100 m
below the surface, the base of the weathered layer is
close to datum, so the elevation correction is small and
changes only slightly with changes in the replacement
velocity. If datum is at the surface, a distance of about
100 m is involved in the elevation correction, and thus
there is a larger difference in the final result, depending
on the replacement velocity used. A 20% error in the
refractor velocity is shown to result in a datum static
correction error of 59 ms.
The above discussion does not necessarily represent
the full picture, as it is just concerned with the computa-

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

244

Static Corrections for Seismic Reflection Surveys

tion using a constant refraction delay time. If the erroneous refractor velocity is also used to derive intercept
times (or delay times) from the original timedistance
data, an increase in these times is observed. When these
are converted to depth, this is likely to offset the
decrease noted above and, in some cases, may result in
an increase in the estimate of the weathering thickness.
For example, at an offset of 200 m, increasing the refractor velocity from 1800 to 2160 m/s increases the delay
time by 18 ms; it is likely that this will represent a greater
depth variation than the 26% referred to above.
Thus, many factors must be considered in evaluating
the errors in datum static corrections derived from
refraction data. In many cases, however, compensating
factors can reduce the final error.

5.8

SHEAR-WAVE SURVEYS

Shear (S-) wave refraction surveys are similar to those


used for compressional (P-) waves, both in terms of
acquisition and interpretation. S-wave velocities are
much slower than P-waves and are almost unaffected
by water saturation (see Section 2.6). In many cases, the
interfaces are different for the two modes of propagation
so that the refractors mapped are likely to be at different
depths. An important example of this is the water table,
which is sometimes mapped with a P-wave survey but
is unlikely to be observed in a S-wave survey.
Various S-wave sources are used; for small-scale surveys, a large wooden plank struck at its ends with a
hammer is often used. The receivers are horizontal geophones, aligned along the line of recording (radial) or
perpendicular to it (transverse) to record SV and SH
data; in practice, three component geophones may be
used. Other aspects of data recording are as described in
Section 5.4 for P-wave surveys. Specifications of the
recording layout must be able to accommodate the Swave velocity and likely refractor depths. Comparisons
of P-wave and S-wave refraction records are shown by
Stmpel et al. (1984), Wiest and Edelmann (1984),
Meissner et al. (1985), and Lawton (1990).
An alternative way of acquiring S-wave refraction
data is to work with converted waves; the most common configuration is recording S-waves from a P-wave
source (e.g., Lash, 1986; Houston et al., 1989; Schafer,
1991). If mode conversion takes place at the interface
between the refractor and weathered layers, at the point
where the head wave or upcoming wave leaves the
refractor (designated PPS), any compressional-wave
refractor can be mapped, such as the water table. Other
mode-converted refractions are designated SPP and
SPS, which are from an S-wave source. The mode-converted PPS arrivals are normally much weaker than the

P-waves so that interference often occurs between the


earlier P-wave arrivals and the S-wave arrivals, even
though horizontal geophones are used to record the
data.
Houston et al. (1989) showed that the difference in
refraction arrival time (T) between the P-wave and the
converted-wave (PPS) arrival was given by

1
T = z
V1S

V12S
1 2
V2

1/ 2

1
V1P

V12P
1 2
V2

1/ 2

, (5.104)

where z is the weathered layer thickness, V1S is the nearsurface S-wave velocity, V1P the near-surface P-wave
velocity, and V2 the refractor velocity (P-wave). If the
higher order terms of V1P/V2 and V1S/V2 are neglected,
equation (5.104) can be simplified (Balachandran, 1975):
1
1
z
T = z

V V1S ) .
+
2 ( 1P
V
V
1S
1P 2V2

(5.105)

If z, V1P , and V2 are estimated from the P-wave survey, V1S can be computed from equation (5.104) or
(5.105). This can then be used to compute the S-wave
time in the weathered layer. However, this is not a full
S-wave datum static correction because it does not
account for the S-wave replacement velocity or the possibility of deeper near-surface layers that should be
included in the computation.

REFERENCES
Ackermann, H.D., Pankratz, L.W., and Dansereau, D.A., 1982,
A comprehensive system for interpreting seismic-refraction arrival-time data using interactive computer methods: U. S. Geol. Surv. Open File Rep. 82-105.
Ackermann, H.D., Pankratz, L.W., and Dansereau, D., 1986,
Resolution of ambiguities of seismic refraction traveltime
curves: Geophysics, 51, 223-235.
Adachi, R., 1954, On a proof of fundamental formula concerning refraction method of geophysical prospecting and
some remarks: Kumamoto J. Sci., 2, 18-23.
Adams, D.C., Miller, K.C., and Baker, M., 1994, Applications
of first break turning-ray tomography to shallow seismic
reflection data processing and interpretation: 64th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
587-590.
Ak, M.A., 1990, An analytical raypath approach to the refraction wavefront method: Geophys. Prosp., 38, 971-982.
Aldridge, D.F., and Oldenburg, D.W., 1992, Refractor imaging
using an automated wavefront reconstruction method:
Geophysics, 57, 378-385.
Ansell, E.A., 1930, Das impulsfeld der praktischen seismik in
graphischer behandlung (Graphical treatment of the
impulsefield in practical seismology): Gerlands
Ergnzungshefte fr argewante Geophysik, 1, 117-136.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Archer, S.H., and Heathcote, C., 1985, Can refraction arrival
times be used to solve the long wavelength statics problem?: 55th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 301-303.
Backus, G. E., and Gilbert, J.F., 1967, Numerical application of
a formulation for geophysical inverse problems: Geophys.
J. Roy. Astr. Soc., 13, 247-276.
Bahorich, M.S., Coruh, C., Robinson, E.S., and Costain, J.K.,
1982, Static corrections on the southeastern Piedmont of
the United States: Geophysics, 47, 1540-1549.
Baixas, F., Canal, P., and Paturet, D., 1985, Investigating
refraction statics methods as a function of geologic complexity: 55th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 290-292.
Baixas, F., and DuPont, R., 1988, Practical view of 3-D refraction statics: 58th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 787-790.
Balachandran, K., 1975, Determination of weathering thickness by a seismic P-S delay technique: Geophysics, 40,
1073-1075.
Banerjee, B., and Gupta, S.K., 1975, The hidden layer problem
in seismic refraction work: Geophys. Prosp., 23, 642-652.
Banta, H.E., 1941, A refraction theory adaptable to seismic
weathering problems: Geophysics, 6, 245-253.
Barbier, M.G., Bondon, P., Mellinger, R., and Viallix, J.R., 1976,
Mini-Sosie for land seismology: Geophys. Prosp., 24, 518527.
Barry, K.M., 1967, Delay time and its application to refraction
profile interpretation, in Musgrave, A.W., Ed., Seismic
refraction prospecting: Soc. Expl. Geophys., 348-361.
Barthelmes, A.J., 1946, Application of continuous profiling to
refraction shooting: Geophysics, 11, 24-42.
Baumgarte, J., 1955, Konstruktive darstellung von seismischen horizonten unter bercksichtigung der strahlenbrechung im raum (Geometrical construction of reflecting
and refracting boundaries derived from seismic time
observations): Geophys. Prosp., 3, 126-162.
Beattie, R., and Wardell, J., 1987, Near-surface modelling by
interactive refraction analysis: Expl. Geophys., 18, 12-13.
Bell, M.L., Lara, R., and Gray, W.C., 1994, Application of turning-ray tomography to the offshore Mississippi Delta: 64th
Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, 1509-1512.
Berge, A.M., and Beskow, B., 1985, A method to determine
the velocities of the seafloor and near-surface sediments:
Geophys. Prosp., 33, 377-399.
Berni, A.J., and Roever, W.L., 1989, Field array performance:
theoretical study of spatially correlated variations in
amplitude coupling and static shift and case study in the
Paris Basin: Geophysics, 54, 451-459.
Bishop, T.N., Bube, K.P., Cutler, R.T., Langan, R.T., Love, P.L.,
Resnick, J.R., Shuey, R.T., Spindler, D.A., and Wyld, H.W.,
1985, Tomographic determination of velocity and depth in
laterally varying media: Geophysics, 50, 903-923.
Reprinted, 1988, in Lines, L. R., Ed., Inversion of geophysical data: Soc. Expl. Geophys., 368-388.
Bratton, R.H., and Musgrave, A.W., 1966, Apparatus for
establishing a common datum plane: United States Patent
3 284 765; (abstract): Geophysics (1967), 32, 545.
Brtz, R., Marschall, R., and Knecht, M., 1987, Signal adjustment of vibroseis and impulsive source data: Geophys.
Prosp., 35, 739-766. Reprinted, 1989, in Geyer, R. L., Ed.,
Vibroseis: Soc. Expl. Geophys., 686-713.

245

Brckl, E.P., 1991, Migration of refraction eventsa combination of traveltime and wavefield techniques: 53rd Mtg.,
Eur. Assn. Expl. Geophys., Abstracts, 36-37.
Brckl, E., and Kohlbeck, F., 1994, The application of stacking
and migration in refraction seismology: 56th Mtg., Eur.
Assn. Expl. Geophys., Paper H022.
Brzostowski, M.A., and McMechan, G.A., 1992, 3-D tomographic imaging of near-surface seismic velocity and
attenuation: Geophysics, 57, 396-403.
Cassinis, R., and Borgonovi, L., 1966, Significance and implication of shingling in refraction records: Geophys. Prosp.,
14, 547-565.
Chon, Y.-T., and Dillon, T.J., Jr., 1986, Tomographic mapping
of the weathered layer: 56th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 593-595.
Chun, J.H., and Jacewitz, C.A., 1980, Automated statics estimation utilizing first-arrival refractions: Presented at the
50th Ann. Internat. Mtg., Soc. Expl. Geophys.; (abstract):
Geophysics (1981), 46, 437.
Chun, J.H., and Jacewitz, C.A., 1981, The weathering statics
problem and first arrival time surfaces: Presented at the
51st Ann. Internat. Mtg., Soc. Expl. Geophys.; (abstract):
Geophysics (1982), 47, 503-504.
Chun, J.H., and Jacewitz, C.A., 1983, First arrival time surface, estimation of statics: Oil and Gas J., Sept. 5, 162-167
and 169-170.
Cooke, D.A., and Schneider, W.A., 1983, Generalized linear
inversion of reflection seismic data: Geophysics, 48, 665676. Reprinted, 1988, in Lines, L. R., Ed., Inversion of geophysical data: Soc. Expl. Geophys., 233-244.
Coppens, F., 1985, First arrival picking on common-offset
trace collections for automatic estimation of static corrections: Geophys. Prosp., 33, 1212-1231.
Coruh, C., Costain, J.K., and Stephenson, D.E., 1993,
Composite refraction-reflection stack sections: 63rd Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
1157-1160.
Cotten, S.A., and Gochioco, L.M., 1991, Investigation of a
buried hillside using seismic refraction: 61st Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 522-525.
Cummings, D., 1979, Computer Program: determination of
depths to an irregular interface in shallow seismic refraction surveys using a pocket calculator: Geophysics, 44,
1987-1998.
Cunningham, A.B., 1974, Refraction data from single-ended
refraction profiles: Geophysics, 39, 292-301.
Cutler, R.T., 1987, Invited introductory paper: seismic topography: 57th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 833-835.
Daly, C., and Diggins, C., 1988, Use of refractor elevation
models in the computation of refraction statics: 58th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
574-577.
Dampney, C.N.G., and Whiteley, R.J., 1980, Velocity determination and error analysis for the seismic refraction
method: Geophys. Prosp., 28, 1-17.
de Amorim, W.N., Hubral, P., and Tygel, M., 1987,
Computing field statics with the help of seismic tomography: Geophys. Prosp., 35, 907-919.
Dent, B., 1989, Surface-consistent 3-D refraction statics: practical advantages and a steep-dip generalization: 59th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
1294-1297.
Dent, B., 1990, On the use of refracted first breaks of marine
reflection data for static and dynamic corrections: 60th

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

246

Static Corrections for Seismic Reflection Surveys

Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded


Abstracts, 1738-1740.
Devaney, A.J., 1984, Geophysical diffraction tomography:
IEEE Trans. Geosci. Remote Sensing, GE-22, 3-13.
Diggins, C., Carvill, C., and Daly, C., 1988, A hybrid refraction algorithm: 58th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 578-581.
Dines, K.A., and Lytle, R.J., 1979, Computerised geophysical
tomography: Proc. IEEE, 67, 1065-1073.
Disher, D.A., and Naquin, P.J., 1970, Statistical automatic statics analysis: Geophysics, 35, 574-585.
Dix, C.H., 1939a, Refraction and reflection of seismic waves I:
Fundamentals: Geophysics, 4, 81-101.
Dix, C.H., 1939b, Refraction and reflection of seismic waves
II: Discussion of the physics of refraction prospecting:
Geophysics, 4, 238-241.
Dix, C.H., 1955, Seismic velocities from surface measurements: Geophysics, 20, 68-86. Reprinted, 1988, in Lines, L.
R., Ed., Inversion of geophysical data: Soc. Expl. Geophys.,
149-167. Reprinted, 1990, in Byun, B. S., Ed., Velocity
analysis on multichannel seismic data: Soc. Expl.
Geophys., 9-27.
Dix, C.H., 1961, The seismic head pulse, reflection and pseudo-reflection pulses: J. Geophys. Res., 66, 2945-2951.
Dix, C.H., 1981, Seismic prospecting for oil: Internat. Human
Res. Dev. Corp.
Dobrin, M.B., 1976, Introduction to geophysical prospecting:
McGraw-Hill Book Co.
Dobrin, M.B., and Savit, C.H., 1988, Introduction to geophysical prospecting: McGraw-Hill Book Co.
Docherty, P., 1992, Solving for the thickness and velocity of
the weathering layer using 2-D refraction tomography:
Geophysics, 57, 1307-1318.
Docherty, P., and Kappius, R., 1993, A workstation implementation of 3-D refraction statics: 63rd Ann. Internat. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, 1166-1169.
Domzalski, W., 1956, Some problems of shallow refraction
investigations: Geophys. Prosp., 4, 140-166.
Donato, R.J., 1964, Amplitude of P-head waves: J. Acoust.
Soc. Am., 36, 19-25.
Donato, R.J., 1965, Measurements on the arrival refracted
from a thin high speed layer: Geophys. Prosp., 13, 387-404.
Duska, L., 1963, A rapid curved-path method for weathering
and drift corrections: Geophysics, 28, 925-947.
Edge, A.B., and Laby, T.H., 1931, The principles and practice
of geophysical prospecting: Cambridge Univ. Press., 339341.
Ervin, C.P., McGinnis, L.D., Otis, R.M., and Hall, M.L., 1983,
Automated analysis of marine refraction data: a computer
algorithm: Geophysics, 48, 582-589.
Evjen, H.M., 1967, Outline of a system of refraction interpretation for monotonic increase of velocity with depth, in
Musgrave, A.W., Ed., Seismic refraction prospecting: Soc.
Expl. Geophys., 290-294.
Ewing, M., and Leet, D.L., 1932, Comparison of two methods
for interpretation of seismic time-distance graphs which
are smooth curves: A.I.M.E. Geophys. Prosp., 263-270.
Ewing, M., Wollard, G.P., and Vine, A.C., 1939, Geophysical
investigations in the emerged and submerged Atlantic
Coastal Plain, Part III, Barnegat Bay, New Jersey, Section:
Bull. Geol. Soc. Am., 50, 257-296.
Farrell, R.C., and Euwema, R.N., 1984a, Refraction statics:
Proc. IEEE, 72, 1316-1329.

Farrell, R.C., and Euwema, R.N., 1984b, Surface-consistent


decomposition of refraction raypaths: 54th Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 424-425.
Farrell, R.C., Koehler, F., and Hawthorne, R., 1981, An automated procedure for refraction statics: Presented at the
51st Ann. Internat. Mtg., Soc. Expl. Geophys.; (abstract):
Geophysics (1982), 47, 503.
Ferree, C.M., and Miller, D.F., 1972, Method and apparatus
for determining near-surface corrections from seismic
data: United States Patent 3 681 749; (abstract): Geophysics
(1973), 38, 203.
Fuller, B.N., and Kusuma, T., 1993, A method for picking first
breaks in 3-component seismic data: 63rd Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 325-328.
Fulton, T.K., and Darr, K.M., 1984, Offset panel: Geophysics,
49, 1140-1152.
Fulton, T.K., and Darr, K.M., 1985, Offset panel: a powerful
analytical tool: Oil and Gas J., May 13, 135-142.
Gamburtsev, G.A., 1946, Correlation refraction shooting:
Geophysics, 11, 59-65.
Gardner, L.W., 1939a, An areal plan of mapping subsurface
structure by refraction shooting: Geophysics, 4, 247-259.
Gardner, L.W., 1939b, Seismograph prospecting: United
States Patent 2 153 920; (abstract): Geophysics, 4, 313.
Gardner, L.W., 1967, Refraction seismograph profile interpretation, in Musgrave, A.W., Ed., Seismic refraction
prospecting: Soc. Expl. Geophys., 338-347.
Garotta, R., 1983, Simultaneous recording of several Vibroseis
seismic lines: 53rd Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 308-310.
Garotta, R., 1984, New approaches to long wavelength residual static corrections: 54th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 678-679.
Garotta, R., 1987, Simultaneous reflection-refraction acquisition: First Break, 5, 327-334.
Gelchinsky, B., and Shtivelman, V., 1983, Automatic picking
of first arrivals and parameterization of traveltime curves:
Geophys. Prosp., 31, 915-928.
Gelius, L.J., Kullerud, A., and Rafto, J.E., 1984, Inversion of
refracted data: an automatic procedure: 54th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
567-570.
Gendzwill, D.J., 1978, A method of weathering correction:
Geophys. Prosp., 26, 525-537.
Gibson, B., and Larner, K., 1984, Predictive deconvolution
and the zero-phase source: Geophysics, 49, 379-397.
Goguel, F.M., 1951, Seismic refraction with variable velocity:
Geophysics, 16, 81-101.
Goulty, N.R., and Brabham, P.J., 1984, Seismic refraction profiling in opencast coal exploration: First Break, 2, No. 5,
26-34.
Grant, F.S., and West, G.F., 1965, Interpretation theory in
applied geophysics: McGraw-Hill Book Co.
Green, R., 1962, The hidden layer problem: Geophys. Prosp.,
10, 166-170.
Green, R., 1974, The seismic refraction methoda review:
Geoexploration, 12, 259-284.
Greenhalgh, S.A, 1977, Comments on The hidden layer
problem in seismic refraction work, Banerjee, B., and
Gupta, S.K., authors: Geophys. Prosp., 25, 179-181.
Greenhalgh, S.A., and King, D.W., 1980, Determination of
velocity-depth distributions by inversion of refraction
time-distance data: Austral. Soc. Expl. Geophys. Bull., 11,
No. 3, 92-98.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Greenhalgh, S.A., and King, D.W., 1981, Curved ray path
interpretation of seismic refraction data: Geophys. Prosp.,
29, 853-882.
Greenhalgh, S.A., King, D.W., and Emerson, D.W., 1980, On
the fitting of velocity functions to seismic refraction data:
Austral. Soc. Expl. Geophys. Bull., 11, No. 3, 78-91.
Hagedoorn, J.G., 1955, Templates for fitting smooth velocity
functions to seismic refraction and reflection data:
Geophys. Prosp., 3, 325-338.
Hagedoorn, J.G., 1959, The plus-minus method of interpreting seismic refraction sections: Geophys. Prosp., 7, 158182.
Hagedoorn, J.G., 1964, The elusive first arrival: Geophysics,
29, 806-813.
Hagiwara, T., and Omote, S., 1939, Land creep at Mt. TyausuYama. (determination of slip plane by seismic prospecting): Bull. Earthquake Res. Instit., Tokyo Univ., 17, 118-137.
Hales, F.W., 1958, An accurate graphical method for interpreting seismic refraction lines: Geophys. Prosp., 6, 285-294.
Hampson, D., and Russell, B., 1984a, First break interpretation using generalized linear inversion: J. Can. Soc. Expl.
Geophys., 20, No. 1, 40-54. Reprinted, 1988, in Lines, L. R.,
Ed., Inversion of geophysical data: Soc. Expl. Geophys.,
96-110.
Hampson, D., and Russell, B., 1984b, First break interpretation using generalized linear inversion: 54th Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 532-534.
Handley, E.J., 1954, Computing weathering corrections for
seismograph shooting: World Oil, 139, No. 6, 118-128.
Hatherly, P.J., 1980, Computer processing of seismic refraction
data: Austral. Soc. Expl. Geophys. Bull., 11, 69-74.
Hatherly, P.J., 1982a, Wave equation modelling for the shallow seismic refraction method: Austral. Soc. Expl.
Geophys. Bull., 13, 26-34.
Hatherly, P.J., 1982b, A computer method for determining
seismic first arrival times: Geophysics, 47, 1431-1436.
Hatherly, P.J., and Neville, M.J., 1986, Experience with the
generalized reciprocal method of seismic refraction interpretation for shallow engineering site investigation:
Geophysics, 51, 255-265.
Hawkins, L.V., 1961, The reciprocal method of routine shallow seismic refraction investigations: Geophysics, 26, 806819.
Hawkins, L.V., and Maggs, D., 1961, Nomograms for determining maximum errors and limiting conditions in seismic refraction surveys with a blind zone problem:
Geophys. Prosp., 9, 526-532.
Hawkins, L.V., and Whiteley, R.J., 1982, Seismic refraction signatures for massive sulfide ore bodies: 52nd Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 401-404.
Heelan, P.A., 1953, On the theory of head waves: Geophysics,
18, 871-893.
Heiland, C.A., 1940, Geophysical exploration: Prentice-Hall,
Inc.
Heldreth, C.G., and OConnell, J.K., 1993, Mississippi Delta 3D seismic data processing case history: 63rd Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 1161-1165.
Hill, N.R., 1987, Downward continuation of refracted arrivals
to determine shallow structure: Geophysics, 52, 1188-1198.
Hollingshead, G.W., and Slater, R.R., 1979, A novel method of
deriving weathering statics from first arrival refractions:
Presented at the 49th Ann. Internat. Mtg., Soc. Expl.
Geophys.; (abstract): Geophysics (1980), 45, 535-536.

247

Hollister, J.C., 1967, A curved path refraction method, in


Musgrave, A.W., Ed., Seismic refraction prospecting: Soc.
Expl. Geophys., 217-230.
Holst, R., Jumper, S., and Pardue, H., 1985, An interactive
seismic refraction statics correction method: 55th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
299-301.
Hoover, G.M., Gallagher, J.G., and Rigdon, H.K., 1984,
Vibrator signals: Proc. IEEE, 72, 1290-1301.
Houston, L.M., Rice, J.A., and Cameron, D.S., 1989, Method
for shear-wave static corrections using converted wave
refractions: 59th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 1298-1301.
Hunter, J.A., Burns, R.A., Good, R.L., and Pullan, S.E., 1991,
Refraction velocity measurements of subsea-bottom sediments using a vertical array in Arctic Ocean waters: 61st
Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, 580-583.
Hunter, J.A., and Pullan, S.E., 1990, A vertical array method
for shallow seismic refraction surveying of the sea floor:
Geophysics, 55, 92-96.
Igarashi, T., and Matumae, Y., 1985, The effectiveness of
stacking refraction method data using a nonexplosive seismic source: 55th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 172-174.
Inoue, S., and Tanaka, K., 1986, Refraction analysis on a personal computer using the principles of the Hales method
and inverse analysis: 56th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 599-602.
Ivansson, S., 1985, A study of methods for tomographic
velocity estimation in the presence of low-velocity zones:
Geophysics, 50, 969-988.
Ivansson, S., 1986, Seismic borehole tomographytheory and
computational methods: Proc. IEEE, 74, 328-338.
Jackson, D.D., 1972, Interpretation of inaccurate, insufficient
and inconsistent data: Geophys. J. Roy. Astr. Soc., 28, 97109. Reprinted, 1986, in Vozoff, K., Ed., Magnetotelluric
methods: Soc. Expl. Geophys., 417-429. Reprinted, 1988, in
Lines, L. R., Ed., Inversion of geophysical data: Soc. Expl.
Geophys., 11-23.
Jakosky, J.J., 1950, Exploration geophysics: Trija Publishing
Co.
Johnson, S.H., 1976, Interpretation of split-spread refraction
data in terms of plane dipping layers: Geophysics, 41, 418424.
Jones, M.G., and Jovanovich, D.B., 1985, A ray inversion
method for refraction analysis: Geophysics, 50, 1701-1720.
Judson, R.D., and Sherwood, J.W.C., 1973, Static corrections
for seismic traces by cross-correlation method: United
States Patent 3 731 269; (abstract): Geophysics, 38, 10021003.
Kaila, K.L., and Narain, H., 1970, Interpretation of seismic
refraction data and the solution of the hidden layer problem: Geophysics, 35, 613-623.
Kaufman, H., 1953, Velocity functions in seismic prospecting:
Geophysics, 18, 289-297.
Kearey, P., and Brooks, M., 1991, An introduction to geophysical exploration: Blackwell Scientific Publications, Inc.
Keliher, J., and Bishop, K., 1993, Overthrust velocity models
from refraction arrivals: 63rd Ann. Internat. Mtg., Soc.
Expl. Geophys., Expanded Abstracts, 825-828.
Ketelsen, K., and Fromm, G., 1982, Interactive static correction procedure with automatic picking of first arrivals:
Presented at the 44th Mtg., Eur. Assn. Expl. Geophys.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

248

Static Corrections for Seismic Reflection Surveys

Ketelsen, K., Marschall, R., and Fromm, G., 1983, Automatic


picking of first-arrivals with two-sided signals (vibroseis):
Presented at the 45th Mtg., Eur. Assn. Expl. Geophys.
Khan, K.A., 1994, An intelligent and efficient approach to
picking first breaks: 56th Mtg., Eur. Assn. Expl. Geophys.,
Paper P155.
Kilty, K.T., Norris, R.A., McLamore, W.R., Hennon, K.P., and
Euge, K., 1986, Seismic refraction at Horse Mesa Dam; an
application of the generalized reciprocal method:
Geophysics, 51, 266-275.
Kirchheimer, F., 1988, 3-D refraction statics by weighted leastsquares inversion: 58th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 794-797.
Knox, W.A., 1967, Multilayer near-surface refraction computations, in Musgrave, A.W., Ed., Seismic refraction
prospecting: Soc. Expl. Geophys., 197-216.
Kumar, G.N., and Raghava, M.S.V., 1981, On the significance
of amplitude studies in shallow refraction seismics:
Geophys. Prosp., 29, 350-362.
Kusuma, T., and Brown, M.M., 1992, Cascade-correlation
learning architecture for first-break picking and automated trace editing: 62nd Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 1136-1139.
Kusuma, T., and Fish, B.C., 1993, Toward more robust neuralnetwork first break and horizon pickers: 63rd Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
238-241.
Laidley, M.E., and Mills, G.F., 1986, Pitfalls in interactive
refraction modeling in severe problem areas: 56th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
597-599.
Landa, E., Keydar, S., and Kravtsov, A., 1995, Determination
of a shallow velocity-depth model from seismic refraction
data by coherence inversion: Geophys. Prosp., 43, 177-190.
Lankston, R.W., 1989, The seismic refraction method: a viable
tool for mapping shallow targets into the 1990s:
Geophysics, 54, 1535-1542.
Lankston, R.W., 1990, High-resolution refraction seismic data
acquisition and interpretation, in Ward, S.H., Ed.,
Geotechnical and environmental geophysics: Soc. Expl.
Geophys., 1, 45-73.
Lankston, R.W., and Lankston, M.M., 1986, Obtaining multilayer reciprocal times through phantoming: Geophysics,
51, 45-49.
Lash, C.C., 1986, P to S conversion by a refracted P-wave:
The Leading Edge, 5, No. 7, 31-34.
Laski, J.D., 1973, Computation of the time-distance curve for
a dipping refractor and velocity increasing with depth in
the overburden: Geophys. Prosp., 21, 366-378.
Laster, S.J., Backus, M.M., and Schell, R., 1967, Analog model
studies of the simple refraction problem, in Musgrave,
A.W., Ed., Seismic refraction prospecting: Soc. Expl.
Geophys., 15-66.
Lavergne, M., 1961, Etude sur modele ultrasonique du problem des couches minces en sismique refraction (Ultrasonic
model study of refracted waves along thin layers):
Geophys. Prosp., 9, 60-73.
Lavergne, M., 1966, Refraction le long des bancs minces rapides et effet decran pour les marqueurs profonds
(Refraction along thin high velocity layers and the screening effect on deep refractors): Geophys. Prosp., 14, 504527.
Lawton, D.C., 1989, Computation of refraction static corrections using first-break traveltime differences: Geophysics,
54, 1289-1296.

Lawton, D.C., 1990, A nine-component refraction statics


experiment: 60th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 1089-1092.
Layat, C., 1967, Modified Gardner delay time and constant distance correlation interpretation, in Musgrave,
A.W., Ed., Seismic refraction prospecting: Soc. Expl.
Geophys., 171-193.
Leet, L.D., 1938, Practical seismology and seismic prospecting, in Mather, K.F., Ed., The century earth science series:
Appleton-Century-Crofts, Inc.
Leven, J.H., and Taylor, F.J., 1989, Cumulative difference statics: method and application: Expl. Geophys., 20, 365-370.
Levin, F.K., and Ingram, J.D., 1962, Head waves from a bed of
finite thickness: Geophysics, 27, 753-765.
Lines, L.R., Ed., 1988, Inversion of geophysical data: Soc.
Expl. Geophys.
Lines, L.R., and Treitel, S., 1984, Tutorial: A review of leastsquares inversion and its application to geophysical problems: Geophys. Prosp., 32, 159-186. Reprinted, 1988, in
Lines, L. R., Ed., Inversion of geophysical data: Soc. Expl.
Geophys., 68-95.
Lo, T.-w., and Inderwiesen, P., 1994, Fundamentals of seismic
tomography: Soc. Expl. Geophys.
Lourantos, G.P., Kirchheimer, F., and Koitka, H., 1991, On the
application of statics in marine 3-D data processing: 53rd
Mtg., Eur. Assn. Expl. Geophys., Abstracts, 90-91.
Luzietti, E.A., Moore, D.E., Smith, G.E., Moldoveanu, N.,
Spradley, M., Brooks, T., and Chang, M., 1995, Innovation
and flexibility: keys to a successful 3-D survey in the transition zone of West Bay Field, Louisiana: The Leading
Edge, 14, 763-772.
Macrides, C.G., and Dennis, L.P., 1994, 2D and 3D refraction
statics via tomographic inversion with under-relaxation:
First Break, 12, 523-537.
MacPhail, M.R., 1967, The midpoint method of interpreting a
refraction survey in Musgrave, A.W., Ed., Seismic refraction prospecting: Soc. Expl. Geophys., 363-415.
Maillet, R., and Bazerque, J., 1931, La prospection sismique
du sous-sol (Seismic exploration of the subsurface):
Annales des Mines, 20, 287-341.
Marsden, D., 1993, Static correctionsa review, Part II: The
Leading Edge, 12, 115-120.
Martin, L.A., 1978, Method of determining weathering corrections in seismic operations: United States Patent 4 101
867; (abstract): Geophysics (1979), 44, 280.
Martin, L.A., and Fenley, W.F., Jr., 1985, Method of determining weathering corrections in seismic operations: United
States Patent 4 498 157; (abstract): Geophysics, 50, 12081209.
McCormack, M.D., and Rock, A.D., 1993, Adaptive network
for automated first break picking of seismic refraction
events and method of operating the same: United States
Patent 5 181 171; (abstract): Geophysics, 58, 912-913.
McCormack, M.D., Zaucha, D.E., and Dushek, D.W., 1993,
First-break refraction event picking and seismic data trace
editing using neural networks: Geophysics, 58, 67-78.
Meeder, C.A., May, J.A., Tinkle, A.R., Wener, K.R., and Huff,
J.F., 1988, Seismic no-data zone, offshore Mississippi Delta:
Part IIImodeling statics corrections: Proc. 20th Offshore
Tech. Conf., 3, Paper 5755, 85-92.
Meissner, R., Stmpel H., and Theilen, F., 1985, Shear wave
studies in shallow sediments, in Dohr, G.P., Ed.,
Handbook of geophysical exploration, Sect. 1. Seismic
exploration. Vol. 15B, Seismic shear waves. Part B,
Applications: Geophysical Press, 224-253.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Merrick, N.P., Odins, J.A., and Greenhalgh, S.A., 1978, A
blind zone solution to the problem of hidden layers within a sequence of horizontal or dipping refractors:
Geophys. Prosp., 26, 703-721.
Morack, J., Hunter, J.A., and MacAulay, H.A., 1984, Iso-offset
time display of seismic refraction data: 54th Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 136-138.
Morgan, N.A., 1967, The use of the continuous seismic profiler to solve hidden layer problems: Geophys. Prosp., 15,
35-43.
Mota, L., 1954, Determination of dip and depths of geological
layers by the seismic refraction method: Geophysics, 19,
242-254.
Mural, M.E., and Rudman, A.J., 1992, Automated first arrival
picking: A neural network approach: Geophys. Prosp., 40,
587-604.
Musgrave, A.W., 1967, Ed., Seismic refraction prospecting:
Soc. Expl. Geophys.
Musgrave, A.W., and Bratton, R.H., 1967, Practical application of Blondeau weathering solution, in Musgrave, A.W.,
Ed., Seismic refraction prospecting: Soc. Expl. Geophys.,
231-246.
Muskat, M., 1933, The theory of refraction shooting: Physics,
4, 14-28.
Musser, J.W., Wason, C.B., and Sudhakar, V., 1991, Using
stacked correlations in time picking for statics: 61st Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
1406-1408.
Naik, S.B.R., Kumar, G.N., and Raghava, M.S.V., 1980, The
correlation refraction method as applied to weathered zone
studies in a granite terrain: Geophys. Prosp., 28, 18-29.
Nettleton, L.L., 1940, Geophysical prospecting for oil:
McGraw-Hill Book Co.
Nolet, G., Ed., 1987, Seismic tomography: D. Reidel Publ. Co.
Northwood, E.J., 1967, Notes on errors in refraction interpretation, in Musgrave, A.W., Ed., Seismic refraction
prospecting: Soc. Expl. Geophys., 459-465.
OBrien, P.N.S., 1955, Model seismologythe critical refraction of elastic waves: Geophysics, 20, 227-242.
OBrien, P.N.S., 1957, The variation with distance of the
amplitude of critically refracted waves: Geophys. Prosp.,
5, 300-316.
OBrien, P.N.S., 1967a, The use of amplitudes in seismic
refraction survey, in Musgrave, A.W., Ed., Seismic refraction prospecting: Soc. Expl. Geophys., 85-118.
OBrien, P.N.S., 1967b, The efficient use of large charges, in
Musgrave, A.W., Ed., Seismic refraction prospecting: Soc.
Expl. Geophys., 152-170.
Olsen, K.B., 1989, A stable and flexible procedure for the
inverse modelling of seismic first arrivals: Geophys.
Prosp., 37, 455-465.
Ongkiehong, L., and Askin, H.J., 1988, Towards the universal
seismic acquisition technique: First Break, 6, 46-63.
Ostrander, W.J., 1992, Method of improving the seismic resolution of geological structures: United States Patent 5 083
297; (abstract): Geophysics, 57, 971.
Paal, E.F., 1987, Method for establishing a surface consistent
correction for the effects of the low velocity layer in seismic data processing: United States Patent 4 695 984.
Pakister, L.C., and Black, R.A., 1957, Exploring for ancient
channels with the refraction seismograph: Geophysics, 22,
32-47.
Palmer, D., 1980, The generalized reciprocal method of seismic refraction interpretation: Soc. Expl. Geophys.

249

Palmer, D., 1981, An introduction to the generalized reciprocal method of seismic refraction interpretation:
Geophysics, 46, 1508-1518.
Palmer, D., 1983, Comment on Curved raypath interpretation of seismic refraction data, Greenhalgh, S.A., and
King, D.W., authors: Geophys. Prosp., 31, 542-543.
Palmer, D., 1986, Refraction Seismics; Handbook of geophysical exploration, Vol. 13: Geophysical Press.
Palmer, D., 1990, The generalized reciprocal methodan
integrated approach to shallow refraction seismology:
Expl. Geophys., 21, 33-44.
Palmer, D., 1991, The resolution of narrow low-velocity zones
with the generalized reciprocal method: Geophys. Prosp.,
39, 1031-1060.
Pant, P.R., and Raghava, M.S.V., 1987, A simple approach to
the masked-layer problem in seismic refraction work:
Geoexploration, 24, 549-556.
Patterson, A.R., 1964, Datum corrections in glacial drift:
Geophysics, 29, 957-967.
Paulson, K.V., and Merdler, S.C., 1968, Automatic seismic
reflection picking: Geophysics, 33, 431-440.
Peraldi, R., 1969, Contribution du traitement numerique a
lanalyse et a linterpretation des enregistrements refraction (Contribution of digital processing to the analysis and
interpretation of refraction records): Geophys. Prosp., 17,
126-164.
Peraldi, R., and Clement, A., 1972, Digital processing of
refraction data study of first arrivals: Geophys. Prosp., 20,
529-548.
Poley, J.P., and Nooteboom, J.J., 1966, Seismic refraction and
screening by thin high-velocity layers: Geophys. Prosp.,
14, 184-203.
Press, F., Oliver, J., and Ewing, M., 1954, Seismic model study
of refractions from a layer of finite thickness: Geophysics,
19, 388-401.
Pritchett, W.C., 1991, An example of simultaneous recording
where necessary signal separation is easily achieved:
Geophysics, 56, 9-17.
Propes, R.L., 1991, Geophysical exploration using near surface structure corrections developed from common endpoint gather stacked traces: United States Patent 5 073 876;
(abstract): Geophysics (1992), 57, 662.
Qin, F., Cai, W., and Schuster, G.T., 1993, Inversion and imaging of refraction data: 63rd Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 613-615.
Raghava, M.S.V., and Kumar, G.N., 1979, The blind zone
problem in multiple refraction layer overburden situations: Geophys. Prosp., 27, 474-479.
Ramananantoandro, R., and Bernitsas, N., 1987, A computer
algorithm for automatic picking of refraction first-arrival
time: Geoexploration, 24, 147-151.
Reynolds, C.B., Reynolds, I.B., and Haneberg, W.C., 1990,
Refraction velocity sections: an aid in shallow reflection
interpretation: 60th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 383-384.
Ricker, N., 1953, The form and laws of propagation of seismic
wavelets: Geophysics, 18, 10-40. Reprinted, 1981, in
Toksz, M. N. and Johnston, D. H., Eds., Seismic wave
attenuation: Soc. Expl. Geophys., 354-384.
Ricker, N.H., 1977, Transient waves in visco-elastic media:
Elsevier Scientific Publ. Co., Inc.
Ristow, D., and Jurczyk, D., 1975, Vibroseis deconvolution:
Geophys. Prosp., 23, 363-379. Reprinted, 1989, in Geyer, R.
L., Ed., Vibroseis: Soc. Expl. Geophys., 620-636.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

250

Static Corrections for Seismic Reflection Surveys

Rockwell, D.W., 1967, A general wavefront method, in


Musgrave, A.W., Ed., Seismic refraction prospecting: Soc.
Expl. Geophys., 363-415.
Rhl, T., 1995, Determination of shallow refractor properties
by 3D-CMP refraction seismic techniques: First Break, 13,
69-77.
Rhl, T., and Lschen, E., 1990, Inversion of first-break traveltime data of deep seismic reflection profiles: Geophys.
Prosp., 38, 247-266.
Russell, B.H., 1988, Introduction to seismic inversion methods: Soc. Expl. Geophys.
Saleh, S.J., 1994, Sensitivity of refraction statics to choice of
the picking range of offsets: 64th Ann. Internat. Mtg., Soc.
Expl. Geophys., Expanded Abstracts, 1525-1528.
Salvatori, H., and Walling, D., 1937, Method of making
weathering corrections in seismic surveying: United States
Patent 2 087 120.
Santosa, F., and Symes, W.W., 1989, An analysis of leastsquares velocity inversion: Soc. Expl. Geophys.
Schafer, A.W., 1991, Determination of converted-wave statics
using P refractions together with SV refractions: 61st Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
1413-1415.
Scheffers, B.C., 1992, Combined traveltime inversion of shallow seismic data: 54th Mtg., Eur. Assn. Expl. Geophys.,
Abstracts, 738-739.
Scheidegger, A.E., and Willmore, P.L., 1957, The use of a least
squares method for the interpretation of data from seismic
surveys: Geophysics, 22, 9-22.
Schenck, F.L., 1967, Refraction solutions and wavefront targeting, in Musgrave, A.W., Ed., Seismic refraction
prospecting: Soc. Expl. Geophys., 416-425.
Schmller, R., 1982, Some aspects of handling velocity inversion and hidden layer problems in seismic refraction
work: Geophys. Prosp., 30, 735-751.
Schneider, W.A., 1971, Developments in seismic data processing and analysis (1968-1970): Geophysics, 36, 1043-1073.
Schneider, W.A., and Kuo, S.-Y., 1985, Refraction modeling
for static corrections: 55th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 295-299.
Selvi, O., 1990, Imaging of more than one refractor by
reverse-time migration: Presented at the 52nd Mtg., Eur.
Assn. Expl. Geophys.
Sheriff, R.E., 1989, Geophysical methods: Prentice-Hall, Inc.
Sheriff, R.E., 1991, Encyclopedic Dictionary of Exploration
Geophysics: Soc. Expl. Geophys.
Sheriff, R.E., and Geldart, L.P., 1982, Exploration seismology,
Vol. 1, History, theory and data acquisition: Cambridge
Univ. Press.
Sheriff, R.E., and Geldart, L.P., 1995, Exploration seismology:
Cambridge Univ. Press.
Silverman, D., 1978, Method of determining weathering corrections in seismic record processing: United States Patent
4 069 471; (abstract): Geophysics, 43, 848.
Simmons, J.L., Jr., and Backus, M.M., 1992, Linearized tomographic inversion of first-arrival times: Geophysics, 57,
1482-1492.
Simmons, J.L., Jr., and Bernitsas, N., 1994, Nonlinear inversion of first-arrival times: 64th Ann. Internat. Mtg., Soc.
Expl. Geophys., Expanded Abstracts, 992-995.
Sims, J., and Mackenzie, B.E., 1973, Computer P-wave picking: Oil and Gas J., Feb. 19, 58-60.
Sjgren, B., 1979, Refractor velocity determinationcause
and nature of some errors: Geophys. Prosp., 27, 507-538.

Sjgren, B., 1980, The law of parallelism in refraction shooting: Geophys. Prosp., 28, 716-743.
Sjgren, B., 1984, Shallow refraction seismics: Chapman and
Hall.
Slichter, L.B., 1932, The theory of the interpretation of seismic
travel-time curves in horizontal structures: Physics, 3, 273295.
Slotnick, M.M., 1950, A graphical method for the interpretation of refraction profile data: Geophysics, 15, 163-180.
Soske, J.L., 1959, The blind zone problem in engineering geophysics: Geophysics, 24, 359-365.
Spagnolini, U., 1991, Adaptive picking of refracted first
arrivals: Geophys. Prosp., 39, 293-312.
Spencer, T.W., 1965, Refraction along a layer: Geophysics, 30,
369-388.
Steeples, D.W., Miller, R.D., and Black, R.A., 1990, Static corrections from shallow-reflection surveys: Geophysics, 55,
769-775.
Stefani, J.P., 1993, Possibilities and limitations of turning ray
tomography: a synthetics study: 63rd Ann. Internat. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, 610-612.
Stevens, J.W., 1982, Determination of static corrections from
refraction travel time: United Kingdom Patent 2 090 405
A.
Stewart, R.R., 1991, Exploration seismic tomography:
Fundamentals: Soc. Expl. Geophys.
Stresau, W.R., Farrell, R.C., and Ford, K.P., 1992a, 3-D refraction analysis and modeling using surface-consistent
decomposition on the work station: 62nd Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 1116-1119.
Stresau, W.R., Hansen, L., Rowe, R.W., and Morales, M.,
1992b, Generalized linear inversion method for near-surface modeling: a 3-D processing case history: 62nd Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
303-306.
Stmpel, H., Khler, S., Meissner, R., and Milkereit, B., 1984,
The use of seismic shear waves and compressional waves
for lithological problems of shallow sediments: Geophys.
Prosp., 32, 662-675.
Swan, H.W., and Booker, A.H., 1984, Accuracy of statics
obtained from single layer refraction modeling: 54th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
422-423.
Taner, M.T., 1986, Near surface imaging by refracted seismic
waves: Presented at the 3rd Ann. Summer Research
Workshop: Soc. Expl. Geophys.
Taner, M.T., Koehler, F., and Sheriff, R.E., 1979, Complex seismic trace analysis: Geophysics, 44, 1041-1063.
Taner, M.T., Lu, L., and Baysal, E., 1988, Unified method for
2-D and 3-D refraction statics with first break picking by
supervised learning: 58th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 772-774.
Taner, M.T., Matsuoka, T., Baysal, E., Lu, L., and Yilmaz, O.,
1992, Imaging with refractive seismic waves: 62nd Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
1132-1135.
Tarrant, L.H., 1956, A rapid method of determining the form
of a seismic refractor from line profile results: Geophys.
Prosp., 4, 131-139.
Telford, W.M., Geldart, L.P., and Sheriff, R.E., 1990, Applied
geophysics: Cambridge Univ. Press.
Telford, W.M., Geldart, L.P., Sheriff, R.E., and Keys, D.A.,
1984, Applied geophysics: Cambridge Univ. Press.
Thompson, J.F., 1963, A technique for solving the low-velocity
layer problem: Geophysics, 28, 869-876.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Thornburgh, H.R., 1930, Wavefront diagrams in seismic interpretation: AAPG Bull., 14, 185-200.
Trostle, M.E., 1967, Some aspects of refraction shooting
through screening layers, in Musgrave, A.W., Ed., Seismic
refraction prospecting: Soc. Expl. Geophys., 469-481.
van Overmeeren, R.A., 1987, The plus-minus method for
rapid field processing by portable computer of seismic
refraction data in multi-layer groundwater studies: First
Break, 5, 83-94.
Veezhinathan, J., Wagner, D. and Ehlers, J., 1991, First break
picking using a neutral network, in Aminzadeh, F. and
Simaan, M., Eds., Expert systems in exploration: Soc. Expl.
Geophys., 179-202.
Vesnaver, A., and Boehm, G., 1994, The contribution of
refracted waves to seismic tomography: 56th Mtg., Eur.
Assn. Expl. Geophys., Paper P062.
Wagner, D.E., Ehlers, J.W., and Veezhinathan, J., 1990, First
break picking using neural networks: 60th Ann. Internat.
Mtg., Soc. Expl. Geophys., Expanded Abstracts, 370-373.
White, D.J., 1989, Two-dimensional seismic refraction tomography: Geophys. J., 97, 223-245.
White, D.J., and Milkereit, B., 1990, Subsurface velocity structure from seismic refraction tomography: 60th Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
1572-1574.
Whiteley, R.J., and Greenhalgh, S.A., 1979, Velocity inversion
and the shallow seismic refraction method:
Geoexploration, 17, 125-141.
Whiteley, R.J., and Palmer, D., 1981, Discussion on
Determination of velocity-depth distributions by inversion of refraction time-distance data, Greenhalgh, S., and
King, D., authors with reply by authors and Emerson,
D.W.,: Austral. Soc. Expl. Geophys. Bull., 12, 52-54.
Wiest, B., and Edelmann, H.A.K., 1984, Static corrections for
shear wave sections: Geophys. Prosp., 32, 1091-1102.
Wiggins, R.A., 1972, The general linear inverse problem:
implication of surface waves and free oscillations for earth
structure: Rev. of Geophys. and Space Phys., 10, 251-285.
Reprinted, 1988, in Lines, L. R., Ed., Inversion of geophysical data: Soc. Expl. Geophys., 10.

251

Wiggins, R.A., Larner, K.L., and Wisecup, R.D., 1976,


Residual static analysis as a general linear inverse problem: Geophysics, 41, 922-938. Reprinted, 1985, Geophysics,
50, 2172-2188. Reprinted, 1988, in Lines, L. R., Ed.,
Inversion of geophysical data: Soc. Expl. Geophys., 24-40.
Willmore, P.L., and Bancroft, A.M., 1960, The time term
approach to refraction seismology: Geophys. J., 3, 419-432.
Wilson, J.L., 1983, Method for determining source and receiver statics in marine seismic exploration: United States
Patent 4 415 997; (abstract): Geophysics (1984), 49, 483.
Won, I.J., and Bevis, M., 1984, The hidden-layer problem
revisited: Geophysics, 49, 2053-2056.
Wong, J., Bregman, N., West, G., and Hurley, P., 1987, Crosshole seismic scanning and tomography: The Leading
Edge, 6, No. 1, 36-41.
Wooley, W.C., Musgrave, A.W., and Gray, H., 1967, A method
of inline refraction profiling, in Musgrave, A.W., Ed.,
Seismic refraction prospecting: Soc. Expl. Geophys., 267289.
Worthington, M.H., 1984, An introduction to geophysical
tomography: First Break, 2, No. 11, 20-26.
Wyrobek, S.M., 1956, Application of delay and intercept
times in the interpretation of multilayer refraction time
distance curves: Geophys. Prosp., 4, 112-130.
Zachariadis, R.G., 1986, Method for determining source and
receiver statics in marine seismic exploration: United
States Patent 4 581 724; (abstract): Geophysics, 51, 1865.
Zanzi, L., 1990, Inversion of refracted arrivals: a few problems: Geophys. Prosp., 38, 339-364.
Zanzi, L., and Carlini, A., 1991, Refraction statics in the
wavenumber domain: Geophysics, 56, 1661-1670.
Zhu, X., and McMechan, G.A., 1988, Estimation of near-surface velocities by tomography: 58th Ann. Internat. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, 1236-1238.
Zhu, X., Sixta, D.P., and Angstman, B.G., 1992, Tomostatics:
Turning-ray tomography + static corrections: The Leading
Edge, 11, No. 12, 15-23.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

252

Static Corrections for Seismic Reflection Surveys

CHAPTER 5 APPENDIX
Glossary of Refraction Interpretation Methods
This glossary of refraction interpretation methods is limited to those published in Sheriffs Encyclopedic Dictionary
of Exploration Geophysics; definitions in quotes are taken directly from Sheriff (1991). In many cases, the reader is
referred to the original reference and to the relevant pages in Sheriff and Geldart (1982). Where appropriate, references
are given to sections in this chapter.

ABC method: A method of computing refractor depth


based on refracted arrivals from sources near the surface. Especially used for weathering thickness determination from sources above the base of the weathering. This method was described in more detail in
Section 5.6.3.

Generalized reciprocal method (GRM): A refraction


interpretation method that is a generalization of the
plus-minus method. It takes into account noncoincidence of the stations used for calculating plus values.
See Palmer (1980). This method was described in
more detail in Section 5.6.5.

Banta method: A curved-raypath correction method


that assumes that successively greater refraction
times have penetrated to greater depths.

Hagedoorn method: Plus-minus method of refraction


interpretation. This method was described in
Section 5.6.4.

Barrys method: A refraction interpretation method


utilizing delay times. See Barry (1967) or Sheriff and
Geldart, v. I (1982, p. 220-221). The delay time
method was described in Section 5.6.6.

Hales (Hales) method: A graphical refraction interpretation method, particularly useful where the
refractor changes depth markedly, such as where
there is considerable relief or over large faults, but
with constant velocity above the refractor. See Hales
(1958) or Sheriff and Geldart, v. 1 (1982, p. 255257).
The technique is based on the common emergent
point from the refractor for both directions of recording and has similarities to the GRM. See also Wooley
et al. (1967), Sjgren (1979), and Inoue and Tanaka
(1986).

Barthelmes method: A refraction interpretation


method involving continuous profiling. See Barthelmes (1946).
Baumgarte ray-stretching method: A graphical refraction interpretation method. Positions of successive
layers are constructed as surfaces tangent to fictitious
wavefronts which are projected backward from the
observing stations. See Baumgarte (1955).
Blondeau method: A method of determining vertical
time to a predetermined depth based on first-break
data and the assumption that the instantaneous
velocity is proportional to a power of the depth. The
modified Blondeau method assumes that the constant of proportionality and the exponent can vary
with depth. See Musgrave and Bratton (1967).
Named for E. E. Blondeau, American geophysicist.
This method was described in more detail in Section
5.6.7.
First-break intercept-time method: A method making
static corrections based on first breaks.
Gardners method: A refraction interpretation method
which involves separating intercept time into constituent delay times associated with the source and
geophone ends of the trajectory. See Gardner
(1939[a]). This technique was described in more
detail in Section 5.6.6.

Intercept method: A method of computing near-surface corrections from the intercept time at zero distance on a timedistance plot of first breaks. This
technique was described in more detail in Section
5.6.1.
Plus-minus method: A refraction interpretation
method using reversed refraction profiles, also called
Hagedoorn method. Let tAB be the surface-to-surface
time between A and B and let tA and tB be arrival
times at various intermediate locations from sources
A and B, respectively. Minus values, tA tB tAB,
are calculated for each location and plotted to give
the velocity of the refractor. Plus values, tA + tB
tAB, are calculated for each location and plotted to
give a picture of the refractors depth. See Hagedoorn
(1959) or Sheriff and Geldart, v. 1 (1982, p. 225). This
method was described in more detail in Section 5.6.4.
Reciprocal method: A refraction method such as the
generalized reciprocal method.

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Chapter 5Refraction Surveys


Slotnick method: A graphical refraction interpretation
method applicable for plane multilayer refractors.
See Slotnick (1950).
Summation method: A method of calculating a
weathering correction to seismic arrival times. For
adjacent interlocked split-dip profiles with sources
just below the low-velocity layer, the correction for
each group is half the sum of the first-arrival times at
that group from the two interlocking records minus
the average high-velocity time between the source
points (obtained by subtracting the uphole time from
the first-arrival time of the group at one of the source
points when a source is fired at the other source
point).
Tarrant method: A graphical refraction interpretation
method applicable where refractor shape varies; see
Sheriff and Geldart, v. 1 (1982, p. 221222) or Tarrant
(1956).
Thornburghs method: A refraction interpretation
method which uses Huygens principle to construct
wavefronts from reciprocal source points, working
backward from the observed arrival times at the surface. The velocities above the refractor must be

253

known for the construction. See wavefront method


and Thornburgh (1930) or Sheriff and Geldart, v. I
(1982, p. 223224). This method was described in
more detail in Section 5.6.8.
Wavefront method: A seismic interpretation method
(often graphical) which involves reconstructing
emerging wavefronts from the arrival times at various geophones from a common source (or the equivalent). The wavefront for the time t is constructed by
striking circles about each geophone position of
radius (t1 t)V, where t1 is the traveltime observed by
that geophone and V is the velocity of the upper
layer. Similar wavefronts can be constructed from
other source points or from the reversed profile of a
refraction interpretation. The solution which locates
the reflector or refractor must satisfy the observed
arrival times. See Rockwell (1967). This method was
described in more detail in Section 5.6.8.
Wyrobek method: A refraction interpretation method
based on applying delay and intercept times to continuous refraction profiling, even where the profiles
are not reversed. See Wyrobek (1956) or Sheriff and
Geldart, v. 1 (1982, p. 196, 222223).

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Downloaded 20 Dec 2011 to 198.3.68.20. Redistribution subject to SEG license or copyright; Terms of Use: http://segdl.org/

Das könnte Ihnen auch gefallen