Geoexploration, 12 (1974) 259-284

o Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

THE SEISMIC REFRACTION METHOD- A REVIEW

RONALD GREEN
Department of Geophysics, University of New England, Armidale, N.S. W. (Australia)
(Accepted for publication December 12, 1973)

ABSTRACT

Green, R., 1974. The seismic refraction method- a review. Geoexploration, 12: 259-284.

The seismic refraction method had its beginning in the war of 1914-18, but it became
established as a prospecting method in the 1920’s and 30’s as a result of successes in the
Gulf area of the United States. The success was an outcome of the parallel development
of improved instrumentation and improved methods of interpretation. Instruments gener-
ally become smaller, lighter, more portable and reliable.
Field technique has steadily improved the signal-to-noise ratio. The interpretation
methods have tended to be simpler to apply and applicable to more realistic geological
field situations.
The seismic refraction method has found new applications in crustal geophysics, recon-
naisance surveying in sedimentary basins, structural engineering, and mining geophysics.
This review brings together the history of the development of the method, a discussion of
the basic theory, field procedures and instrumentational developments and a discussion
of methods of interpretation. Some examples are given as an illustration.
It is suggested that students will find the review of value in having an account of the
history, the theory and applications in the one paper, and professional geophysicists will
find it of interest and of value in indicating the continued developments that have taken
place over the years. It would seem to indicate that future developments will be towards
lighter equipment with improved information gathering capabilities coupled with auto-
matic and portable data processors which will interpret the data directly in terms of real-
istic geological structures.

INTRODUCTION

The seismic refraction method is applied to situations in which there are

essentially flat-lying strata, each having differing propagation velocities for
compressional waves. A second requirement is that the thickness of each
layer is small compared with the length of the refraction spread, i.e., the
distance from the shot-point to the most distant detector. Fig.1 shows the
usual arrangement. Note that the thickness of the strata is small compared
with the spread length X(SG,). However, there are other requirements for
the successful conduct of a refraction survey, which will be merely stated in
this paper. Principally, there must be few discontinuities in the interfaces be-
tween the strata. One such discontinuity (marked F, fault) is shown in Fig.1.
For the disturbance to travel out from the shot-point to the detectors along
260

dry sand 200

alluvium 600

marls 2000
Shot point 6 detectors (geophones)

sandstone 2500

lImestone 4000

igneous mlrusive 5000

Fig.1. The usual arrangement for refraction surveying. Note that the spread length is long
compared with the layer thickness. A normal fault (8’) in the lower layer is shown.

Fig.2. A layering that could be investigated by the refraction method. The velocities in
m/set are given for the different rock types.

ray-paths such as shown in Fig.1, a necessary condition is that the seismic

velocity increase with depth, i.e., u, > un_ I > . . .ul . Such a condition may
appear to be unduly restrictive but in most cases the condition holds. A plau-
sible explanation is that in the expression for the velocity u, of the compres-
sional seismic wave, u = [ (h + 4~/3) / p ] ‘/* , the elastic constants k and p, in-
crease at a faster rate with compaction than does the density p, thereby in-
creasing the velocity with depth. The increase of velocity with depth is spoken
of as “normal”. As a rough guide, numerical values of velocity have been
listed against rock types in Fig.2.
Because the velocity of the seismic ray through the rock strata is the phys-
ical property being measured, it is obvious that if accuracy is to be main-
tained, both the horizontal distances and the timing must be accurate. For
local surveys the required accuracy is 100mm and lmsec. With modern mea-
suring and timing equipment such accuracies are easily obtained. Difficulties
in interpretation arise largely from the complexity that can occur in geologi-
cal structures.

APPLICATIONS

The refraction method can be used for many types of structural problems.
These may involve spread lengths from tens of metres to hundreds of kilo-
metres. The difference is not in principle but one of scale. For example, re-
fraction surveys have been carried out with spreads of up to 1OOOkm to de-
termine the thickness of the crust above the Mohorovicic discontinuity.
These surveys have been carried out both at sea (Officer and Ewing, 1954)
and on land (Pakiser et al., 1960). On a somewhat reduced scale, the seismic
refraction method has been used as a reconnaisance tool in relatively geo-
logically unknown areas (Layat et al., 1961; Blundun, 1956; Bartelmes, 1946)
so as to hold down the costs of prospecting in areas of high operating cost. As
noted also in the above paper, in areas of rough topography and in areas of
 261

limestone deposits where solution-caving occurs, the refraction method can

have advantages over the seismic reflection method in providing more accurate
depth determinations. One of the most comprehensive accounts of the seis-
mic refraction method is given in a volume edited by Musgrave (1967). The
references given are particularly useful. However, the emphasis of the volume
is on applications to regional geology and in the search for oil. Consequently,
the application of the refraction method to such problems will not be dis-
cussed further, because of the limits of space and a desire to confine the
paper to exploration and engineering applications (Hawkins, 1963).
In this field, the refraction method has been most successful when applied
to shallow depth (< 100m) problems. For example, reliable results are ob-
tained for the following types of investigation:
(1) The location of bedrock depression in coastal dune area, as possible
sources of restricted amounts of sweet water (Bonini and Hickok, 1958).
(2) The location of ancient buried river channels when looking for placer
deposits (Edge and Laby, 1931).
(3) Measurement of the thickness of overburden for proposed road, pipe-
line and quarry locations.
(4) To determine the composition of rock from the seismic velocities and
hence the possibility of removing the rock by bulldozer, ripper or explosive
(Moore, 1952).
(5) To determine the interface depths and rock types for foundations or
structures such as buildings, bridges, tunnels and dams (Stam, 1962).

EARLY HISTORY OF SEISMIC REFRACTION

Fundamental and practical investigation into the propagation of seismic

waves were carried out by Rayleigh (1885) and Love (1911), but it was not
until World War I that the technology was developed to provide for the ac-
curate registration and timing at a number of receivers of the arrival of the
airwave from the cannon shot and thereby locate the position of the cannon.
Dr L. Mintrop and Mr E.V. McCollum served on opposing sides during the
war in carrying out this work! However, geophysical prospecting using a
form of the refraction method can be considered fairly to have begun in 1924
with the discovery of the Orchard dome in the Gulf of Mexico coastlands by
the Gulf Production Co. By 1925 the method was well established but in the
early days nothing in the way of scientific papers was published until the
pioneering publications of Heiland (1929) and Barton (1929) appeared. A
comprehensive account of the refraction method was given by Muscat (1933).
As often happens, it was not until much later (De Golyer, 1935; Weatherby,
1940) that historical accounts of the early days of seismic refraction pros-
pecting were documented.
It should be pointed out that initially the pragmatic method known as
“fan-shooting” was used and salt domes, intruded into flat-lying sediments,
were detected by the presence of early arrivals resulting from the higher velo-
city of seismic waves through rock salt (%5,5OOm/sec).
Both the higher velocity of the salt dome and the flatness of the sediments
were necessary for the method to give satisfactory results. Ascan be seen in
Fig.3 the normal time-distance plot shows the refracted arrivals distinctly.
Typical early arrivals are also shown. Fig.3 also shows the features that are
to assume importance in later refraction work: (1) the apparent decrease in
refractor velocity with increasing down-dip; and (2) the presence of a vertical
boundary (between the salt and the sediments) being indicated by a change
in slope of the time-distance plot (Fig.3).

distance ’

Fig.3. The time-distance plot of a refraction survey carried out over a salt dome intruded
into gently dipping sediments.

The use of reversed profiles, the field procedures and methods of interpre-
tation in line with current practice had appeared by 1961. Layat et al. (1961)
describes the use of large spreads for the determination of the depths to re-
fractors in the Sahara and the paper is representative of the use of the refrac-
tion method for reconnaisance surveys.
There was a lack of application of the refraction method to engineering and
shallow prospecting surveys until the late 1950’s, but the equipment in use at
the time, being designed for reflection work, was unsatisfactory for engineer-
ing applications because of the equipment’s bulk, as well as its expense of
operation. This equipment was replaced by simple, portable equipment in-
troduced by Gough (1952), and further developed by Mooney and Kaasa
(1958). Some idea of the rapid development of simple, single-channel por-
table seismic refraction equipment can be gauged from a paper by Stam
 263

(1962) who used such an instrument to determine the thickness of overbur-

den in the Manicouagan river, Quebec. Similar surveys were carried out by
Linehan and Murphy (1952). Just as the simple single-channel seismic equip-
ment had revolutionized refraction work on the land, the seismic sparker
and gas exploder introduced by Knott and Hersey (1956) revolutionized re-
fraction work in bays and estuaries (McGuinness et al., 1962; Hobson, 1970;
Allen, 1972).
The situation today is that the equipment both for land and water use is
appropriate on the basis of weight, reliability, ease of operation and accuracy.
Special feature and requirements of shallow seismic equipment will be dis-
cussed in further detail later (see p. 276 ff.).

CLASSICAL INTERPRETATION

There are many ways in which the methods of interpretation of seismic

refraction results may be discussed. It seems to be that the most suitable,
from the point of view of ease of understanding, and the acquisition of the
ability to apply the methods to data from real situations, is to begin by
studying idealized situations, which are usually referred to as “classical struc-
tures”, and after familiarization with these structures to consider the manner
in which the models have to be modified to accommodate the difficulties
that arise from failures in the models. These failures are caused by the depar-
ture, in a significant degree, of the real structures as encountered in the field
from the idealized structures.
It has already been pointed out in the introduction that the refraction
method is to be used when the spread-length is long compared with the layer
thickness, and provided the layer thicknesses are large compared with the
dominant wavelengths of the propagating wave, the seismic phenomena may
be discussed in terms of geometric ray paths and there is little need to intro-
duce the complexity of wave theory (Ewing, Jardetzky and Press, 1957). The
early work by Slotnick (1950) is based on ray theory and for most refraction
problems the treatment is adequate.
As the first of the idealized structures, consider the case of a single iso-

time
I

Shot pant ground surface

/ velocity “( lnterfoce
velocity v,

Fig.4. The time-distance plot over a single horizontal layer.

264

tropic horizontal layer of thickness, z, and velocity, u, overlying an isotropic

half-space (Fig.4).

Single layer case

From Fig.4 by simple geometry, it can be seen that the travel time for the
direct ray is given by:

t, = x/u1 (1)
The slope of the line is given by:

dt, jdx = l/u, (2)

Hence, ul can be found.
For the refracted ray:

t2 = x/u, + 2(2, /u, ) cos i12 (3)

where:

k, = arcsin (u, /u, ) (4)

The slope of the line is given by:

dt, /dx = l/v, (5)

and hence it can be found.

Let the direct ray and the refracted ray intersect at the point (x 12, t,* ).
It is obvious that for 0 < x < xl2 the direct ray, tl arrives before the re-
fracted ray, i.e.: t, < t2 ; when x < x,~ but tl > t2 ; when x > xl2 and:

x = x,~ ; when t, = t12 = t2. (6)

The distance, x 12 is termed the critical distance and t12, the critical time.
The time intercept on the ordinate, ti by the refracted ray is given by setting
x = 0 in eq.3, i.e.:

ti = 2(z1 /vi ) cos il2 (7)

Hence z1 can be found.

Alternately :

2@, h 1~0s il2 = t12 - xl2 IQ (8)

which also determines z.
While in theory the determination of the depth z may be made by utilizing
either the point xl2 or the intercept time ti, when experimental errors which
affect the determination of the slope of the refracted ray l/u2 are considered
(Steinhart and Meyer, 1961), it is preferable to use the critical distance xi2
to determine z (Zirbel, 1954). The preference for the critical distance has
been pointed out also by Meidav (1960).
 265

Two layer case

The above discussion can be extended to two horizontal layers over a uni-
form half-space. From Fig.5, it can be seen that:
t1 = X/VI (9)

For the first refracted ray:

tz = x/v, + 2 (2, /VI ) cos i1* (10)
For the second refracted ray:
t3 = x/u, + 2 (z2 /u2 ) cos iZ3 + 2 (2, /u, ) cos iI (11)
From eq.6, v1 = xlZ/xIZ (12)
and
v2 = (x23 - x12 ) /(t23 - tl2) (13)
From eq.10:
t 12 = x12 /v2 + 2 (zl /VI) cos iI2 (14)
hence .zl is obtained.
From eq.11:
l/v3 = dt3 /dx (15)
and hence:
t 23 = xz3 /v3 + 2(zZ/v2 ) cos iz3 + 2(2, /VI ) cos il3 06)
and because z2 is the only unknown in eq.16 the depth z2 can be found.

Shot pomt ground surtace

velocity v3

Fig.5. The time-distance plot over two horizontal layers.

266

If there are (n- 1) layers, the time distance equation for the (n- l)th refrac-
tor is given by:
t, =X/V, +2(2,-1/u,-,)~~~in-l,n +2(zn-z/~n-2)~~si,-2,n f (17)
n- 1
= X/Q -t, 2 x Zh/Uk Cos ikn (18)
k=l

It is obvious that:

uk = @k+l,k -xk,k-l)/@k+l,k -fk,k-1) (191

and:
l/v, = dt, /dx (26)
hence the (n- l)th value for the depth zk can be found.
In cases where the velocity increases with depth it is often easier to use the
empirical expression such as: u = cz Irn ; 4 < n < 33 than to consider a large
number of discreet layers (Banta, 1941; Wyrobek, 1959; Acheson, 1963).

Horizontal layers are a very special case, and a more general case for con-
sideration is dipping layers, where @his the dip to the west of the kth inter-
face. To begin with, consider a single interface such as is shown in Fig.6, and
using the nomenclature as given:
t 2 LL= 22, cos ilzlul +x sin (ilz - 91 )lu, (21)
and:
t 2d = 2~ cos i12& +x sin (il2 + o1 ) /ctt (22)
Note that the apparent velocities, up-dip and down-dip, from the refractor,
viz., dt,,/dx and dt,d/dx, are different:

u2u = vi /sin (ilz - 4l ) (23)

u2d = u, /sin (ilz + 41 ) (24)
It is alSO obvious that u2U >U*d
By simple algebra:
il2 = (l/2) (arCSin vl /vzd + amin U1h2d) (25)
and
 267

ground surface ‘2

Fig.6. The time-distance plot over dipping planar layers.

The true refractor velocity, u2 is given by:

v2 =2cosq3-(uUv~)/(V, + W) (27)
It is important to note that the direct ray always gives the velocity of the
upper layer, u1 ; at the deeper end of the spread the critical distance x 12u > x 12d ;
and also note that the travel time between two points is always the same
irrespective of the direction of travel of the ray. That is, the time from S1 to
S, is the same as from SZ to Si . A graphical method of solution of the dip-
ping layer problem was developed by Slotnick (1950).
For multiple dipping layers the algebra becomes heavy and furthermore,
because actual field cases are usually not worked out on the basis of the dip-
ping layer formula, it is proposed not to derive the expressions for multiple
dipping layers. However, for those who may an academic interest in the prob-
lem, a clear exposition has been given by Heiland (1946) and Dooley (1952).

Elevation and weathering corrections

Topographical irregularities have to be taken into consideration. Consider

the case as shown in Fig.7. It can be seen that the elevation correction, Ate to
be added to the recorded travel-time, is given by:
At, = (2d + h - e, - eg )/uO cos iI (28)
In Fig.7 a weathered zone is shown. Let u. be the velocity in the weathered
zone of thickness z under the geophone detector. If allowance for the weath-
ered layer is to be made, the expression for Ate has to be modified by the
addition of the time taken to pass through the weathered layer of thickness
z0 , i.e.:
At, = (2d + t + h - e, -eg) / u. cos iI (29)
268

hillock
deprewon

/ Yeloclty “2

thickness t
Fig.7. Topographic irregularities and the corrections which have to be applied because of
a weathered layer whose velocity is u,

It is usually necessary to carry out a restricted survey to determine the thick-

ness of the weathered layer, zO, along the profile.

Wavefront diagrams

Up till now the interpretation of refraction results has been based on time-
distance plots obtained from summing the time taken along each of the legs
of the geometrical ray path from shot-point to receiver. Nevertheless, dating
from the pioneering days of Thornburg (1930) there has been an approach
to interpretation which seeks to present a picture of the advance of the propa-
gating wavefront. The point is made by the advocates of the wavefront
methods that a wavefront has a physical and observable reality, whereas a ray
path is an abstract concept.
As can be seen from Fig.8 the successive wavefront positions (spacing
s - uAt) present a clear physical picture of the propagation of the seismic
waves but the chief and fatal difficulty was that a succession of wave fronts
is laborious to construct. Hagedoorn (1959) revived the wavefront method
but it did not gain wide acceptance mainly because of the large amount of
computation required. It is possible that the use of a high-speed computer
and plotter could revitalize the method. One such method has been proposed
by Ocola (1972), but it is more applicable to seismic crustal studies than
seismic prospecting. Nevertheless Hagedoorn’s interest in breaking away
from geometric constructions based on plane surfaces for interfaces did lead
indirectly to renewed interest in a method introduced initially by Edge and
Laby (1931) and referred to by them as “the method of differences”.
Let td be the travel time from a to b (Fig.9). Hence, provided u2 >> u, :

t,b = zah’, + x&/us + zb/u, (39)

 269

time

distance

Shot pomt
ground

Fig.%. The successive wavefronts when a single layer lies above a basal refraction.

Fig.9. The field set-up for the applications of “the method of differences” (Edge and Laby,
1931). The shot-points are at (I and c and the detector at b.

tcb = &f+-‘, + X,b/Vz + zbul (31)

tix = ZaUt + (Xnb + xac)/& + z&h (32)
so:
(tab + tcb - ta,) VI = 2zb (33)
The velocity u1 is determined from a separate observation of the direct
ray. It is to be noted that the assumption is made that there is significant dif-
ference between the two velocities (u2 >> ul ) such as occurs where there are
placer deposits over a high velocity basement. Of particular importance is the
method’s ability to handle an irregular basement. It represents a departure
from the assumption of planar interfaces between the strata. Because of this
it more closely represents the actual field situation.
270

The delay-time methods

This method has been developed in response to the usual field situation,
in which a refraction crew is called to work (Gardner, 1939). As is shown in
Fig.10, let there be a number of layers that are separated by nonplanar sur-
faces. This problem was first examined by Barthelmes (1946) and Wyrobek
(1956).

‘13

Fig.10. The ray paths and the application of “delay-time method” to non-planar surfaces.

Let the first shot be fired by S, and let the travel-time of the refracted
waves to the geophones (of which G is one representative geophone), and
also to an additional geophone placed at Sz , be recorded. Let the second shot
be fired at S2 and the travel-time of the refracted wave to the geophones
~including one placed at S2 ) be recorded also.
The significant times are: t(SI G), t(S1 Sz ) = t(S, 52%) and t(S, G); by simple
algebra (Hawkins, 1961):
t(S1 G) + t(S, G) - t(S, S, ) = 22, cos ilz ,‘ul (34)
But the left-hand side of equation 34 is the delay-time (or “reciprocal” time
according to Hawkins, (1961), and can be read off the time-distance plots.
The surface layer velocity, v1 can be read off directly and the critical angle,
i,2, determined from cos iI2 = cos (arcsin (vi luz ) ) (35)
is not strongly dependent on an accurate value of v2, which can be obtained
also from the time-distance plot. I have found that only a small error is in-
troduced from inaccuracies in the estimate for uz.
Writing At for the delay time, the perpendicular distance from G to the
 271

interface is given by:

.zg = At. u1 / (cos i,?) (36)
It is important to note that it is necessary that the two arrivals from shot-
points S1 and S2 be refracted arrivals. It should be noted that even if the
geophone g is placed on a small mound the position of the interface is still
located at its correct depth below the geophone, zg. If the arrivals are not re-
fracted arrivals from the same refractor, the position of the shotpoints must
be moved to ensure that it is so. Moving the shot-point out causes the ray
to be refracted from a deeper layer.
In the general form, the expression to determine the thickness of the nth
layer, zn is:
rl- I
z, =(At, -c z/2 cos ik,n f 1 ) un/cos i?z,n + 1 (37)
k=l

The above method supersedes the early methods based on the principle of
the delay times of reverse shots that were developed for single refractors by
Tar-rant (1956) and Hales (1958).

Basement irregularities

In the classical approach, geological faults are indicated by the travel times
of reversed profiles (Fig.11). In moving to the right away from the shot-point,
the travel time plot is the same as for any single layer case, until the fault is
reached. At distance well beyond the fault the travel time is the one appro-

time

Shotpoint, S2

d2

Fig.11. A reversed profile shot over a fault.

priate to a single layer of increased thickness, and the travel-time is given by:
t, = 2d* cos ii2 /VI + x/u, + (d, - dt ) cos i12lu, (33)
The increase in time, At gives the throw of the fault, (& -dl ), because:
At = (&-d1 )/u, cos iI (39)
In the reverse shooting the travel-time plot is as shown in Fig.11. Note
that at distances beyond the up-throw of the fault, the arrivals come early,
but again:
At = (dz-dl )/ul cos iI2 (40)
thereby checking the previous determinations of the amount and direction
of the throw of the fault.

Vertical interface

It is most important to carry out reversed profiles as a routine procedure.

Consider the case (Fig.12) of a vertical interface between two rock types
with velocities ul and u2. If the shot is fired as S1 and the detectors laid out
to the right, the travel time will be shown by curve (a). This curve has the ap-
pearance of a layer of velocity u1 , overlain on a refractor of velocity u2. How-
ever, the reversed shot would have a travel time as shown by curve (b).
Because it is impossible to have a high velocity layer on top of a low velo-

Itime

distance

Fig.12. The effect of shooting over a vertical interface. Note the slope of the time-
distance curves.
 273

city refractor, the presence of the interface between two rock types whose
seismic velocities are u1 and uz , is clearly indicated.

DIFFICULTIES WITH THE SIMPLE MODELS

A familiarity with the simple classical models would suggest that the seismic
refraction method would be highly successful in clearly delineating struc-
tures. However, there are a number of difficulties which arise, but which
are often not specifically commented upon. We will now examine a number
of these difficulties.

Injection 0 f energy

For refraction surveys on land, the two most common energy sources are
gelignite explosions or an electrical discharge or “sparker”. (McGuinness et
al., 1962; Allen, 1972). A review of other energy sources is given by Wardell
(1970).
It is obvious that the explosives release the greater amount of energy, but
both energy sources deliver an impulse to the ground which has a sharp
leading edge, and is suitable for determining accurately the time of arrival of
the seismic wave of each of the detectors (Sharpe, 1942).
The form of the seismic spectrum is shown in Fig.13. However, there are
a number of factors which profoundly modify the spectrum (O’Doherty and
Anstey, 1971).

\,
\
Amplitude \
\
log (A) \
\
(m/set/ Hz) \
\
\
1

log ($1 Hz

Fig.13. The seismic spectrum showing the effect of increased charge size.
 Increasing the charge size increases the energy released, but the dominant
frequency is also shifted. Increasing the charge size lowers position of the
dominant frequency (Nicholls, 1962) and, as shown by Hamilton (1972), the
lower frequencies are more strongly attenuated. Another variable that has a
much more serious effect on the spectrum is the type of ground. Granite is
much richer in the transmission of higher frequencies than sandstone. On one
occasion I was required to carry out a seismic refractor survey over an area,
the surface of which was covered with fill made up of cinders and rubbish.
With very modest charges (100 g), large craters were blown in the soft fill and
what energy was imparted to the ground, was in the form of low frequency
waves which was useless in providing a sharp arrival at the detectors. In-
creasing the size of the charge did not give any sharper arrivals but served
solely to blow large craters (Linehan and Murphy, 1962). In other words
the seismic method could not provide ~tisfacto~ results. On the other hand,
if the charge can be fired under water there is very efficient coupling between
the explosive and the water, and between the water and the earth. In other
words, smaller charges can be used when the firing is carried out in water.
Arons and Yennie (1948) state that 25% of the energy is radiated from an
underwater shot.
If charges are fired on the surface, there is poor coupling to the ground -
most of the energy passes directly into the atmosphere. However, Buffet and
Layat (1960) have reported that more energy is imparted to the ground if a
group of charges is suspended a metre or so above the ground rather than
placed directly on the ground.
There are advantages of increased energy injection to be gained by burying
the charge and this can be improved further by tamping the charge with
water.
The depth of burial should be sufficient to ensure that the charge does not
blow out in the time taken for the first half cycle of the first refracted arrival,
(0.01 set). For this reason and for safety a buried charge should be detonated
from the upper part of the charge.
The essential requirement of a seismic source is that its spectrum contains
sufficient energy at a frequency which matches the band-pass frequency of
the seismic detecting and recording equipment.

Picking apt arrival

The amplitude of the arrival is the most useful property in picking the ar-
rival. Frequency plays very little part in identifying the arrival (Hagedoorn,
1964) but it is most important that the frequency response of the detecting
and recording equipment match the spectrum of the waveform generated by
the explosion.
In seismic refraction surveying the attenuation of the seismic wave with
distance is not severe being at 0.1 dblwavelength for rocks whose velocity is
of the order of 3,000 m/set.
 275

Poorly consolidated and poorly sorted sediments have a higher attenuation

rate, largely brought about by scattering of seismic energy. Consequently, as
the distance from the shotpoint to the detector increases, the higher frequen-
cies in the waveform are selectively attenuated and sharp arrivals are more
difficult to identify. With regard to identifying arrivals, it is always the first
arrivals which are most clearly identified. Later arrivals hidden in the coda
disturbance following the first arrivals are much more difficult to identify.
If the upper layer is a very wet soil (velocity u1 ), the velocity of sound in
air (334 m/set) can be greater than the velocity in the soil, in which c=e the
first arrival when the detector is near the shotpoint, can be the airwave. Care
should be taken whenever the first arrival has a velocity close to the velocity
of sound in air, to determine the path of the arrival. (Mooney and Kaasa,
1962).
The correct identification of the first arrival is greatly improved by main-
taining the background noise as low as possible. Chief sources of noise are wind
and cultural noises. Because the amplitude of these noises varies with time,
the opportunity is often taken of firing and recording during a time of low
noise.

Difficult seismic conditions

It has been pointed out that the seismic refraction method requires that
the velocity of each layer increase with depth. However, such is not always
the case. Pakiser and Black (1957) recorded a high velocity mudstone plate
within the lower velocity Ahinarump Conglomerate.
If a layer of substantial thickness has a velocity greater than the velocity
of deeper layers, no refracted arrivals from the deeper layers will be received
(Thralls and Mossman, 1952), and consequently the method provides no in-
formation about the deeper layers. However, if the high velocity layer is thin,
it is found that refracted arrivals from deeper layers do occur. This is be-
cause ray theory does not strictly hold when the layer is thin compared to
the seismic wavelengths, and on the basis of physical wave theory, significant
energy will be propagated through the thin layer (Levin and Ingram, 1962).
Provided the high velocity layer is a thin plate, the high frequency compo-
nent of the waveform is preferentially radiated as the disturbance propagates
down the plate and consequently at a distance there is only the low frequency
component present. However, a low frequency component does not record
well and the first detectable arrival is the wavelet that has been refracted
from a layer below the high velocity plate. In addition to thin rock layers be-
having in this way, ice layers, and in urban areas concrete layers, function as
high-velocity plates.

Hidden layers

Let there be two horizontal layers overlying a half-space and having thick-
ness and velocities as shown in Fig.14.
2’i6

Fig.14. The structure which can lead to a hidden layer and the time-distance plot illus-
trating the refracted arrivals from the intermediate layer, in all cases a secondary arrival.

Even though u3 > u2 > ul, it was shown by Shima (1957) and Soska (1959)
that it was possible that the refracted arrival from the top of the intermediate
layer may never be a first arrival.
Green (1962) showed that the intermediate layer was hidden if the thick-
ness of the top layer exceeded a critical thickness given by:
z1 (mm) = (t13-~13/u1)~1/(2 cos il3) (41)
Fig.14 shows a typical situation and travel-time plot where a hidden layer
occurs.
Morgan (1967) pointed out that if the reflection record from a continuous
seismic profiling system were used to indicate the presence or absence of an
intermediate layer, the information could be used to determine the thickness
.zl and z2 unambiguously.
The position concerning the hidden layer problem can be summarized by
saying that if the possibility of a hidden layer is not considered and if it does
occur in a given area, the calculated depths will have no validity whatsoever.
However, it is easy to examine the seismic record and see if late refracted ar-
rivals are occurring, and if they are, to allow for the thickness of the hidden
layer. In other words the difficulty associated with the hidden layer can be
overcome easily.

METHODS OF IMPROVING THE PERFORMANCE OF A SEISMIC REFRACTION

SYSTEM

Instrumentation

The early shallow refraction instruments introduced by Gough (1952) were

simply timers in which a gate was opened by the hammer blow or explosion
and closed by the arrival of the refracted seismic wave. The major difficulty
with the instrument was in adjusting its gain so that, on one hand, the gate
was not closed prematurely by noise nor on the other hand, failed to be
closed because of the lack of amplitude of the seismic arrival.
There was a considerable improvement gained with the introduction of
instruments which had an adjustable variable time delay and the seismic ar-
rival was displayed on a cathode-ray tube so that the operator could be as-
sured by a visual check that the arrival was the genuine seismic arrival. Repeat-
ability of the recorded waveform would assure the operator of the genuine-
ness of the seismic arrival. Even so, the instrument performed poorly in areas
of high noise because clear arrivals and repeatable waveform could not be ob-
tained because of the low signal/noise ratio. Cultural and atmospheric noise
(Fig.15) occupies the same frequency band as the seismic signal (Frantti,
1963). Consequently, there is improvement to be made by restricting the pass
band of the instrument to the dominant frequencies in the seismic signal.
However, the pass band cannot be too restricted otherwise ringing in the out-
put will occur. Also, if the pass band is narrow, sharp first breaks could not
be recorded and such breaks are necessary for accurate timing of the seismic
arrivals. The manner in which the amplitude and dominant frequency changes
in response to charge size and type of ground has been discussed.
Nevertheless, an improvement can be made because it is possible to use the

I
O-l I
I I. I

100
frequenc:0 Hz

Fig.15. The spectrum of cultural and atmospheric noise. The figure is a generalization of
the results of Frantti (1963).

Fig.16. The improvement in the signal/noise ratio by using an integrating seismic recorder.
Note the growth of the record with repeated impacts. The correct travel time is 43 m-set
as shown in the final frame.
278

property of the repeatability of the true seismic signal and the statistical ran-
domness of the noise. Consequently, if the received signal consisting naturally
of both signal and noise is generated by repeated hammer blows or explosions
and the individual signals are added together, the true signal will be in phase
on every occasion and will constructively add together, whereas the random
noise will destructively add. By the use of repeated signals the signal/noise
ratio can be improved as the half power of the number of signals sent. It is a
most effective method of improving the resolution of a seismic arrival. An
integrating seismograph is marketed by the Bison Instruments Inc. Fig.16
shows the improvement in signal/noise with repeated impacts. The instrument
permits seismic refraction work to be carried out satisfactorily in a high noise
area, in which it was previously impossible to operate with advantage.
An excellent review of the older seismic refraction instruments is given in
a paper by Hobson (1970).

Interpretation

In the interpretation of seismic refraction data it is advisable to use the

delay-time m&ho& (Hawkins, 1961) rather than the more restrictive models
dealing with dipping multiple planar layers (Heiland, 1946; Dooley, 1952).
As pointed out previously the depth to the nth interface is given by:

11v,/cos
in,n+,
n-1
2, = atn- C Z/gCOSik,n+ (42)
L k=l

where it,, is the time delay. While the calculation is routine, it is rather tedi-
ous when it has to be carried out by hand or in a step-by-step operation on a
calculator.
In the past, extensive use of nomograms has been made (Meidav, 1960;
Knox, 1958), and while their use does eliminate much of the numerical eal-
culations, it is more convenient to use a portable pro~ammable calculator
for the reduction of seismic data. Portable programmable calculators such as
Hewlett- Packard 9800 series model 20, permit the entire operation to be run
with the operator supplying only the layer velocities initially and the individ-
ual delay times and distances for each station for the depths to the interfaces
below that station to be determined. It should be pointed out that, while the
advent of the programmable calculator has made it possible to calculate ex-
peditiously the depths to many layers, practical experience has shown that if
more than three layers are incorporated in the solution the depths to the
lower interfaces become unreliable (error 50%) because of the accumulation
of errors, as is obvious from the iterations involved in the expression for z,
in eq.42. Hirschleber (1971) has pointed out that, by using a seismic array
and obtaining multicoverage it is sufficient to record shots in one direction
only and thereby simplify the field technique. He also suggests that the digital
 279

processing of the records have considerable advantage over the interpretation

of analogue records.

Field operations in areas of multiple layers

The success of any geophysical survey largely depends upon an adequate

field operation. Consider a two layer case, and let the object of the survey be
to determine the depths to the two interfaces from G (Fig.17).

“3

Fig.17 To obtain recognizable arrivals refracted from each of the interfaces it is necessary
to have the shot-points at a number of distances, such as shown. Sl to S6 are the shot-
points and G is the single geophone. It is also possible to interchange shotpoints and geo-
phone, i.e., to have one shot-point and 6 geophones.

It is necessary to have at least 3 shot-points to the right of the geophone at

G so that the direct ray and also the refracted arrivals from each of the inter-
faces be received. It is furthermore desirable that a second set of arrivals be
obtained to indicate approximately the velocity of each layer and to assist in
verifying the reliability of the picked arrivals.
In addition, to apply the delay time methods it is also necessary to have a
similar set of shot-points on the left hand side of the geophone. Consequent-
ly, this increases the number of shot-points to twelve for depths to two inter-
faces. Once familiarity with the area is achieved the number of shot-points may
be halved.
The spacing between adjacent detector stations is controlled by the detail
required. Closer spacing obviously increases the amount of work to cover a
given traverse length but if the interfaces are irregular, and remembering that
it is mandatory to have at least two stations per average wavelength in the
irregularities in the basement, then there is no alternative but to carry out
the survey at a closer spacing (Hirschleber, 1971). If the depth at one station
can be reasonably predicted from the depth measurements made at adjacent
stations, then the sampling density is adequate.

FIELD EXAMPLES

Beach sands

The first example is from Sorrell, Tasmania. The problem was to locate
280

sweet ground water to provide a limited supply for a toilet block at an iso-
lated swimming beach. It was estimated that the demand would be lo-20 m3
of water per week, with the demand principally at the weekends. The geo-
logical situation was that the beach and extensive dune sands covered a wave-
cut platform now at a few metres below present-day sealevel. The structure
is indicated in Fig.18. Of particular significance are the depressions in the
wavecut platform. These depressions are structures which offered the possi-
bility of holding usable accumulations of water. This is especially true be-
cause many of the depressions are filled with boulders which allow rapid
movement of water, thereby providing a copious supply of water once the
accumulation of boulders is penetrated. The problem therefore is to map the
surface of the wave-cut platform which is at a depth of 3-4 m below the sur-
face of the sand-dune. The annual rainfall is about 600 mm/yr, which is more
than adequate to maintain fresh water in the rock depressions and to prevent
accumulations of salt water. No surface fresh water is to be found on the
sand-dunes.
SIW Of sweet water
bore beach

sand

basement

Fig.18. A sand covered wave-cut platform which can provide structures capable of holding
water accumulations.

The velocity of the seismic wave through the sand is 900 m/set and the
velocity through the basement is 2,500 m/set. Because of the significant velo-
city contrast between the two materials it is easy to distinguish between the
direct and refracted arrivals. The seismic problem can be considered as a
single layer over a slightly irregular basement.
The result over a depression can be seen in Fig.19. At the 135 and 140 pegs
a depression about 2 m deep and 5 m wide can be seen.
The results from parallel survey lines show that the depression is elongated
towards the inland dunes but closed towards the sea. Consequently the struc-
ture can provide a storage for water seeping from the inshore dunes and
furthermore, it is unlikely that the water will be contaminated by sea-water.
The sweetness of the water may be determined by means of a resistivity sur-
vey carried out over the depressions in the wave-cut platform, the resistivity
of salt water being very much lower than the resistivity of fresh water (Unz,
1959).
Similar problems have been discussed in the paper by Eaton and Watkins
(1970). However, with the depth of the wave-cut platform being only 3-4 m
 281

Surface

v, -730-850 m/sac

79m

“2’ 1930-2440 m/set

2t.Om

distance pegs

s s _
z c :: yy F! _
-
7 z
10 s8 _ %
L 8
7 0
L

Fig. 19. Time-distance plot over a depression in a wavecut platform.

Fig.20. The detail information obtained from seismic refraction lines shot over the site of
the proposed Royal Mint, Canberra (after Hawkins, 1963). The mint is now built.

below the sand cover it was more convenient to use an auger and to drill and
sample the water directly, especially as there was a high probability of the
water being fresh in any chosen depression.
No quantitative pumping tests have been carried out but the water supply
has proved adequate to meet all demands to date. It can meet the sharp
demand at weekends and storage in relation to inflow is sufficient to meet
the overall demand.

Clay deposits

The question to be determined was the amount of clay in a river valley ad-
jacent to a proposed dam site. The clay had formed from the weathering of
metamorphic rocks (schists) and the clay was needed for the construction of
an impervious membrane in a rock-filled dam wall. The clay ranged in depth
from zero to four metres on proceeding from the edge of the valley to its
centre. A number of drill holes were put down, but the seismic refraction
method showed that the interface between the clay and the underlying schist
maintained reasonably constant depth. By means of a combination of the
drill hole information and the seismic refraction data, the actual amount of
clay that was available was accurately determined.

The Royal Mint, Canberra

In this seismic investigation (Hawkins, 1963) not only are the interfaces
identified between the surface layer of unconsolidated sediments, the inter-
mediate layer of weathered/broken rock, and the unweathered rock, but
from the seismic velocities, the physical properties of the rocks are indicated.
Fig.20 is given to indicate the range of velocities and depths recorded. If the
empirical relation of Brown and Robertshaw (1953) is used, it can be seen
that the intermediate layer is weathered rock that can be ripped easily by
bulldozers. The seismic data has also shown the presence of a vertical low
velocity zone in the unweathered rock and this has been interpreted as a zone
of fracturing. It is probable that such a zone may have been missed if only
a drilling program had been used.
It can be said that the use of both drilling and seismic work provided a
much more complete coverage of the site at only slightly increased cost com-
pared with that which could be achieved if either method had been used in
isolation.

CONCLUSION

It has been shown that from a bold and primitive start the seismic refrac-
tion method has developed specialized techniques for crustal geophysics,
regional geological work and various small scale investigation problems of a
geological or engineering nature.
Methods of interpretation have been developed which are simple to use
and are accurate. Portable programmable calculators greatly facilitate the in-
terpretation of refraction data. The equipment necessary for seismic refrac-
tion surveys has also been developed to the stage where it is light and por-
table, easy to use, and reliable. It can be used in high noise area. Because of
the rapidity of coverage with the refraction method, the cost of employing the
method is low. When employed with a limited drilling program, continuous profiles
of precise data can be obtained at a fraction of the cost of an entire drilling
program.

REFERENCES

Acheson, C.H., 1963. Time depth and velocity-depth relations in Western Canada. Geo-
physics, 28: 894-909.
Allen, F.T., 1972. Some characteristics of marine sparker seismic data. Geophysics, 37:
462-470.
Arons, A.B. and Yennie, D.R., 1948. Energy partition in underwater explosion phenomena.
Rev. Mod. Phys., 20: 519-536.
Banta, H.E., 1941. A refraction theory adaptable to seismic weathering problems. Geo-
physics, 6: 245-253.
Barthelmes, A.J., 1946. Application of continuous profiling to refraction shooting. Geo-
physics, 11: 24-4 2.
Barton, D.C., 1929. The seismic method of mapping geological structure. A.I.M.E. Geo-
phys. Prosp., 81: 572-624.
Blundun, G.J., 1956. The refraction seismograph in Alberta foothills. Geophysics, 21:
828-838.
 283

Bonini, W.E. and Hickok, EA., 1958. Seismic refraction method in ground water exptora-
tion. A.I.M.E. Trans., 211: 485-488.
Brown, P.D. and Robertshaw, J., 1953. The in-situ measurement of Young’s modulus for
rock by a dynamic method. Geotechnique, 3: 283.
Buffet, A. and Layat, C., 1960. Nouvel aspect de la refraction seismique en Sahara: tirs
non enter&. Geophys. Prosp., 8: 45-67.
De Golyer, E., 1935. Notes on the early history of applied geophysics in the petroleum
industry. J. Sot. Pet. Geophys., 6: l-10.
Dooley, J.C., 1952. Calculation of depth and dip of several layers by the refraction seismic
method. Au& Bur. Min .Res. Bull., 19: 27-35.
Eaton, G.P. and Watkins, J.S., 1970. The use of seismic refraction and gravity methods in
hydrological investigations. In: L.W. Morley (Editor), Mining and Ground Water Geo-
physics/ 1967. Dep. Energy, Mines, Resour., Can., Econ. Geol. Rep., 26: 544-568.
Edge, A.B. and Laby, T.H., 1931. Principles and Practice of Geophysical Prospecting.
Macmillan, London, pp. 339-340.
Ewing, M., Jardetzky, W.S. and Press, F., 1957. Elastic Waves in Layered Media. McGraw-Hill,
New York, N.Y., 380 pp.
Frantti, G.E., 1963. The nature of high frequency earth noise spectra. Geophysics, 28:
547-562.
Gardner, L.W., 1939. An aereal plan of mapping a subsurface structure by refraction
shooting. Geophysics, 4: 247-259.
Gough, D.I., 1952. A new instrument for seismic exploration at very short ranges. Geo-
physics, 15: 81-101.
Green, R., 1962. The hidden Iayer problem. Geophys. Prosp., 10: 166-170.
Hagedoorn, J.G., 1959. The plus-minus method of interpreting seismic refraction sections.
Geophys. Prosp., 7: 158-182.
Hagedoorn, J.G., 1964. Tbe elusive first arrival. Geophysics, 29: 806-813.
Hales, F.W., 1958. An accurate graphical method of interpreting seismic refraction lines.
Geophys. Prosp., 6: 285-294.
Hamilton, EL., 1972. Compressional-wave attenuation in marine sediments. Geophysics,
37: 620-646.
Hawkins, V.L., 1963. Seismic investigations on the foundation conditions at the Royal
Mint site, Canberra. Proc. R. Sot., N.S.W., 96: 133-139.
Hawkins, L.V., 1961. The reciprocal method of routine shallow seismic refraction inves-
tigations. Geophysics, 26: 806-819.
Heiland, CA., 1929. Modern instruments and methods of seismic prospecting. A.I.M.E.
Geophys. Prosp., 81: 625-653.
Heiland, C.A., 1946. Geophysical Exploration. Prentice-Hall, New York, N.Y., 1013 pp.
Hirschleber, U., 1971. Multicover measurements in refractor shooting. Geophys. Prosp.,
19: 345-356.
Hobson, G.D., 1970. Seismic methods in mining and ground water explorations, In: L.W.
Morley (Editor), Mining and Ground Water Geophysics / 1967. Dep. Energy, Mines,
Resour., Con., Econ. Geol. Rep., 26: 148-176.
Knot, S.T. and Hersey, J.B., 1956. High resolution echo sounding techniques and their
use in bathymetry. Deep Sea Res., 4: 36-44.
Knox, W.A., 1958. A slide rule for near-surface refraction problems. Geophysics, 23:
154-163.
Layat, C., Clement, A., Pommier, G. and Buffet, A., 1961. Some technical aspects of re-
fraction seismic prospecting in the Sahara. Geophysics, 26: 437-446.
Levin, F.K. and Ingram, J.D., 1962. Head waves from a layer of finite thickness. Geo-
physics, 27: 753-765.
Linehan, D. and Murphy, V.J., 1962. Engineering seismofogy applications in metropolitan
areas. Geophysics, 27: 213-220.
Love, A.E.H., 1911. Some Problems of Geodynamics. Cambridge Univ. Press, Cambridge,
180 pp.
284

McGuinness, W.T., Beckman, W.C. and Officer, C.B., 1962. The application of various
geophysical techniques to specialized engineering projects. Geophysics, 27 : 221-236.
Meidav, T., 1960. Nomograms to speed up seismic refraction computations. Geophysics,
25: 1035-1053.
Mooney, H.M. and Kaasa, R.A., 1958. New refraction seismograph. Rev. Sci. Instr., 29: 290-
294.
Mooney, H.M. and Kaasa, R.A., 1962. Air waves in engineering seismology. Geophys.
Prosp., 10: 84-92.
Moore, R.W., 1952. Geophysical methods adapted to highway engineering problems. Geo-
physics, 17 : 505-530.
Morgan, N.A., 1967. The use of the continuous profiler to solve hidden layer problems.
Geophys. Prosp., 15: 35-43.
Muscat, M., 1933. The theory of refraction shooting. Physics, 4: 14-38.
Musgrave, A.W. (Editor), 1967. Seismic Refraction Prospecting. Sot. Explor. Geophys.,
Tulsa, Okla., 604 pp.
Nicholis, HR., 1962. Compiling explosive energy to rock. Geophysics, 27: 305-316.
Ocola, L.C., 1972. A non-linear least-squares method for seismic refraction mapping. Geo-
physics, 37: 260-287.
O’Doherty, R.F. and Anstey, N.A., 1971. Reflections on amplitudes. Geophys. Prosp.,lS:
430-458.
Officer, C.B. and Ewing, M., 1954. Geophysical investigations in the emerged and sub-
merged Atlantic coastal plain. Bull. Geol. Sot. Am., 65: 653-670.
Pakiser, L.C. and Black, R.A., 1957. Exploring for ancient channels with the refraction
seismograph. Geophysics, 22: 32-47.
Pakiser, L.C., Press, F. and Kane, M.F., 1960. Geophysical investigation of Mono Basin,
California. Bull. Geol. Sot. Am., 71: 415-448.
Rayleigh, Lord, (Strutt, John William,) 1885. On waves propagated along a plane surface
of an elastic solid. Proc. Lond. Math Sot., 17: 4-11.
Eharpe, J.A., 1942. The production of elastic waves by explosion pressures. Geophysics,
7,144-154; 311-344.
Shima, E., 1957. Note on depth calculations by seismic refraction method. J. Sot. Explor.
Geophys., Jap., 10: 16-18.
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.
Stam, J.C., 1962. Modern developments in shallow seismic refraction techniques. Geo-
physics, 27: 198-212.
Steinhart, J.S. and Meyer, R.P., 1961. Minimum statistical uncertainty of the seismic re-
fraction profile. Geophysics, 26: 574-587.
Tarrant, L.M., 1956. A rapid method of determining the form of seismic refraction from
line profile results. Geophys. Prosp., 4: 131-139.
Thornburg, H.R., 1930. Wavefront diagrams in seismic interpretation. Bull. Am. Assoc.
Prosp. Geophys., 185-200.
Thralls, H.M. and Mossman, R.W., 1952. Relation of seismic corrections to surface geology.
Geophysics, 17: 218-228.
Unz, M., 1959. Interpretation methods for geophysical exploration of reservoirs. Geo-
physics, 24: 109-141.
Wardell, J., 1970. A comparison of land seismic sources. Geoexploration, 8: 205-229.
Weatherby, B.B., 1940. The history and development of seismic prospecting. Geophysics,
5: 215-230.
Wyrobek, S.M., 1956. Application of delay and intercept times in the interpretation of
multi-layer refraction time distance curves. Geophys. Prosp., 4: 1X2-130.
Zirbel, N.N., 1954. Comparison of break-point and time intercept methods in refraction
calculations. Geophysics, 19: 716-721.