(Wissenschaft Und Kultur 27) Markus Båth (Auth.) - Introduction To Seismology (1979, Birkhäuser Basel) PDF
(Wissenschaft Und Kultur 27) Markus Båth (Auth.) - Introduction To Seismology (1979, Birkhäuser Basel) PDF
(Wissenschaft Und Kultur 27) Markus Båth (Auth.) - Introduction To Seismology (1979, Birkhäuser Basel) PDF
~ ~ ~ ~ ~ ~ ~ ~ ~ w ~ _ ~ ~ _ ~ ~ _ _ ~ ~ ~ ~
., .. .., . J) .", 0
'" " " , J '"i l j'"
lti ; \\ ('
,." o f •
o
\J, ."
,
\ t:
.. ~ " ,'~~,t -J ~ ..
' /.
.- -l..
., ~ " .,
~.'l r' 'L \
'. ~.-::
(~~~ , }/.~
",
" ;.~ , '< ,,~ \ ' .;'; i "'· , I, • ~ I •
~
., j' 'l'~.">,.; ~ \.; ;{~', '.: ,. ,,.', '\
» ,,'1' ~ .\ . . " ~ " ,<. " " " f ~~ ..
.~~ 'C'; ' ;r, , ~'
. (
.,
'i_'}'''''~ "
. ",- ,.
" , '} '" ~
'"
. ,~
h 10 .... . . - .. :~
\~~.~~~ .. " . ' . ..,
'~-t, . '....
0'/
." ,::~_,.
,
....
. J)
'.
; ....
,
.
,
.'h
,
.. ,
/
....
. J)
... ,..,
"\
. "
... 1 '-81
T,'
o
.~t" ~ ~ ~ ':" ~ ~ ~M ~ " ~ ~ ~ ~ I~ ~* ' ~ Jo .J., 'SS) '';' '1~ .Ja .~ .&10 .d .c .:., .~,f!>.., ~ .;C .• .m .u, J~
World seismici.y map compiled from ESSA, Coast and Geode.ic Survey, epicen". data, 1961-7 Depths 000-700 km
[aner M. BARAZANC' and J. DORMAN, Bull . S<ism . Soc. Amer. 59. 369- 380 ( 1969)1.
Markus Bath
Introduction
to Seismology
Second, Revised Edition
Bath, Markus:
Introduction to seismology / Markus Bath. - 2.,
rev. ed.
(Wissenschaft und Kultur; Bd. 27)
Einheitssacht.: Introduktion till seismologin (dt.)
ISBN 978-3-0348-5285-2 ISBN 978-3-0348-5283-8 (eBook)
DOI 10.1007/978-3-0348-5283-8
In reviewing the first English edition of this book it has been con-
sidered important not to change its scope, level or purpose, but rather
to bring all information up-to-date. This concerns partly tabular material
on large and destructive earthquakes (Chapter 5), partly data on nuclear
explosions (Chapter 11). Fields in which recent significant developments
have taken place have called for added sections. This concerns earthquake
prediction (Chapter 10) and planetary, especially moon, seismology
(Chapter 12). The seismology curriculum (Chapter 13) has been replaced
by a somewhat changed, newer version, and the Literature Review has
been up-dated by a selection of recent literature. Beyond this, only minor
additions and modifications, corrections of a few misprints, etc have
been done.
Chapter I
the term 'geophysics' to denote what has been listed under 'solid-earth
physics'. Geochemistry is another related subject, which has not been listed
separately here, as it forms a part of several of the subjects in Figure I.
GEOCOSMOPHYSICS
NOSPHERICRESEARCH
GEOMAGNETISM
METEOROLOGY
GEOELECTRICITY
GEOSCIENCE TECTONOPHYSICS
GRAVIMETRY
GEOCOSMOGONY
GEOCHRONOLOGY
Fig. 1. Subdivision of geoscience, geophysics and solid-earth physics into different disciplines. For other
subjects listed the subdivisions are left out.
England in 1849. This is the first time that we come across the waves P
and S, now so well known in seismology. POISSON also studied the free
radial vibrations of a solid sphere, a problem which was also studied later
in England by LAMB and others.
In 1845, STOKES observed that the elastic resistance which an isotropic
body exerts to an applied stress can be split into a resistance against com-
pression or tension and a resistance against shear (by 'isotropic' we mean
that the elastic properties are independent of direction). Compression or
tension is due to normal stress, whereas shear is due to tangential stress.
STOKES also defined two parameters or elasticity moduli to express these
resistances. These two parameters are nowadays called the modulus of
compressibility and the modulus of rigidity.
We have already mentioned POISSON'S discovery of longitudinal and
transverse waves through the interior of a solid elastic body. In 1887, Lord
RAYLEIGH discovered still another type of elastic wave, which instead
propagates along the surface of a body. The velocity or speed of propa-
gation of this wave type was found to be lower than for the two first men-
tioned. These surface waves are also well known in seismology and are
called Rayleigh waves, after their theoretical discoverer. Another important
type of seismic surface wave, i.e. Love waves after the Englishman LOVE,
was not found until in 1911.
The German scientist A. SCHMIDT published in 1888 a paper in
which he discussed the propagation of waves through the earth's interior.
He emphasized that in general the wave velocity must increase with depth
in the earth and that because of this, the wave paths will be curved and
not rectilinear. The curved wave paths will have to be concave towards
the earth's surface. About the same time, KNOTT in England investigated
the energy of reflected and refracted waves.
Among other problems, which were studied at an early stage of elasticity
theory, we mention also conditions on impacts between two solid bodies.
The early elasticity research workers devoted most of their efforts to basic
research. However, their results have later been of the greatest significance
in a number of applied branches, such as technology, seismology, etc.
Thus, we have seen that the main types of seismic waves, nowadays regularly
found on our seismograph records, had been discovered by mathematicians
long before any seismic records had been obtained. For easier review,
Development of the Theory of Elasticity 19
generally have a magnification which is too high to give really good records
of the largest shocks.
Soon after the initial studies of the first seismograms, the identifica-
tion of longitudinal and transverse waves and of Rayleigh waves was clear.
Thus, in the 1890's OLDHAM in England and WIECHERT in Germany asserted
independently that longitudinal and transverse waves and surface waves
exist in the seismograms. Through this discovery the link with elasticity
theory was established, and this initiated a very fruitful period in seismology.
The decades around the turn of the century are characterized by the new
combination between earthquake observations and the elasticity theory.
Efforts to explore the earth's internal constitution by means of seismic
records were pursued with much enthusiasm. On the whole, it can be
maintained, that in this stage the research workers were more inclined to
suspect sharp discontinuity surfaces in the interior, as soon as some observa-
tion seemed to call for such a surface. Later, several such discontinuity
surfaces disappeared, i.e. they were not necessary to explain the observa-
tions, and they were for the most part replaced by zones of more gradual
transition. However, in the course of time, the pendulum swings, and
recently mantle discontinuities begin to appear again. One discontinuity
surface which has remained in spite of all reinterpretations and has been
strongly confirmed is the sharp limit of the earth's core. Its depth was
placed at 2900 km by GUTENBERG in 1913, a depth value which later in-
vestigations have not been able to modify considerably. During the 1930's
the inner core was discovered, having its surface or rather zone of transi-
tion at a depth of around 5000 km. Another well-known and well-established
discontinuity is the base of the earth's crust. This was found in 1909 by
A. MOHOROVICIC by means of seismic records of earthquakes in Croatia.
After him, this surface is called the Mohorovicic discontinuity or, for
brevity, the Moho. The epoch 1910-1940 is characterized by an intensified
study of earthquakes and the earth's interior by means of seismograms.
At the end of this period, seismology had arrived at fairly accurate values
of the internal properties of the earth, such as the elastic wave velocities
(calculated from observed travel times of P and S waves), and moreover
the distribution within the earth of density, pressure, gravity and elastic
parameters. Another discovery in this period (by the English seismologist
TURNER in 1922 and the Japanese seismologist WADATI in 1928) concern-
26 Scope and History of Seismology
Fig. 2. Schematic relation between output and input in earlier (A) and later
INPUT (B) stages of the development of seismology.
Chapter 2
Seismographs
0) b) c) d)
(Fig. 3b), the recording pen traces a displacement downwards. The cor-
responding holds for a horizontal displacement in the case of the horizontal
pendulum (Fig. 3d). These considerations are of fundamental importance
in the transcription of a given record into the corresponding ground motion.
In principle, the pendulums in Figure 3 are seismographs, but of
a very unsatisfactory construction. There are three main objections which
can be made against them:
1. Damping. The pendulums in Figure 3 are practically undamped.
2. Period. The formula for the free period of a pendulum tells us
that pendulums of the construction shown in Figure 3 can only have short
periods, if the size of the instruments should be kept within reasonable
limits.
3. Magnification. The pendulums in Figure 3 work with practically
no magnification of the ground motion.
In the following, we shall see how we can eliminate these short-
comings.
Fundamental Principles 31
11
b)
0)
8
b)
}
w
Fig. 5. Examples of pendulum suspensions
through which arbitrarily long periods can be
achieved: a) vertical pendulum (W=pendulum
mass, A = horizontal rotational axis, B = suspen~
sion point for the helical spring F), b) horizon-
tal pendulum (W=pendulum mass, A and B=
suspeI15ion points, AB = rotational axis).
vertical, then the pendulum boom and the pendulum are in equilibrium
in any position, in other words the free period becomes infinitely great.
By tilting AB, a fixed equilibrium position and a finite period are achieved.
The more AB is tilted, the shorter the period will be and the greater the
restoring force, i.e. the force that restores the pendulum to its equilibrium
position after deflection.
Transmission of pendulum motion to record. Magnification. Practically
all translation seismographs presently in use are pendulum instruments.
This means that the basic principle is the same for all of them. Nonetheless,
different types may deviate from each other by the manner in which the
pendulum motion relative to its frame is transmitted to the record. We
can distinguish two main types for this transmission:
1. Direct-recording seismographs.
2. Electromagnetic seismographs.
In seismographs of older date (for instance, the Wiechert) the motion
is transferred from the pendulum to the recording pens in a purely mecha-
nical way and simultaneously magnified (Fig. 6). In addition, the record-
ing is mechanical (stylus on smoked paper). These things entail that the
seismograph becomes rather bulky and that friction is introduced both
in the mechanical transmission and above all in the mechanical recording.
Fundamental Principles 35
" w
E
~
Fig. 7. The principle of a mechanical-optical seis-
mograph. W = pendulum (horizontal), L = light
source, C=slit, E=lenses, H=mirrOT, R=recording
drum.
In this and in the following section we shall learn how the basic ideas
described above have been applied in seismograph constructions. In fact,
there are often a great variety of constructions which agree in principle,
and therefore we shall limit the discussion to the most representative and
most commonly used types.
Mechanical seismographs. We have already mentioned the Wiechert
seismograph, which is one of the best-known types (Fig. 6). As an example,
I might mention the Wiechert seismograph that was installed at Uppsala,
Sweden, back in 1904. It has a pendulum of 1000 kg and it records the
two horizontal components (E and N) but not the vertical component.
This seismograph is still in operation, but not only out of pious considera-
tions. By its relatively low magnification (maximum around 240 times at
a period of about 8-9 sec), the instrument is still the very best one for
the largest earthquakes. For these the motions are so strong that the records
of the modern, highly sensitive instruments become almost unreadable.
Pendulum Seismographs 39
Fig. 8. The suspension (T) of the stationary mass (pendulum W) in the Wood-Andersoll
T- torsion seismograph. In this case the pendulum mass is of the order of a few grams.
The height of the instrument is usually around 36 em.
\~
F .......
I
Fig. 12. Cross-section of the transducer according to H.
P BENIOFF. M = magnet. 0 = pole-pieces, N = armature, P =
13cm coils. At zero position the four air gaps are each 2 mm wide.
the armature, the air gaps between Nand 0 are varied and thus the magnetic
reluctance (=magnetic resistance) is varied with a corresponding varia-
tion of the magnetic flux through the magnetic circuit. In this way, an
electromotive force is induced in the coils P, and this is then recorded via
a galvanometer. The part shown in Figure 12 is called a transducer and
is in effect nothing other than a modification of an ordinary telephone
receiver.
In a usual construction of BENIOFF'S reluctance seismograph, the trans-
ducer contains eight coils. Of these, four are connected in parallel to a
short-period galvanometer (usually of free period 0.2 or 0.7 sec), while
the other four coils are connected in series to a long-period galvanometer
(usually of a free period of 80-100 sec). In this way, it is possible to pick
out two quite different spectral ranges from one and the same seismometer.
In the Benioff seismometers, pendulum masses around 100 kg are used.
The pendulum (Fig. 11) is restricted in its relative motion to a vertical
direction by flat springs above and below the pendulum mass. By altering
the tension in these springs it is possible to vary the free period of the pen-
Pendulum Seismographs 45
work, as well as the need for darkrooms for recording and developing,
instruments with visible recording have been introduced in recent years.
They have the further advantage of permitting an immediate inspection
of the records. There are several different types available on the market,
such as ink-writing recorders, hot-stylus recorders, etc. In order to operate
these, an amplifier has to be installed between the seismometer and the
recorder. Therefore, the operation may be more expensive and less depend-
able than the photographic one.
Quite another principle for the recording is applied in the so-called
digital seismographs, developed at some institutes in the USA, among
others. In modern research large electronic computers are constantly used_
As a rule, the ordinary records cannot be used directly with computers,
but have to be digitized first (i.e. the curve has to be represented at equal
intervals by numerical values, corresponding to the continuous curve)_
This transcription of the record is frequently a laborious job, at least when
more sophisticated digitizers are not available. Therefore, in the digital
seismographs this trouble is eliminated and one may get directly a series
of numbers (or some other equivalent data form) instead of the continuous
record. Another newer development, which also facilitates the application
of computers, consists in the recording of seismometer outputs on mag-
netic tape.
Calibration. Frequently, electromagnetic seismographs are adjusted to
critical (aperiodic) damping, both for the seismometer and the galvano-
Pendulum Seismographs 47
meter. In this way the picture of the ground motion will be as true as pos-
sible, and moreover, the mathematical treatment will be considerably
simplified. In modern electromagnetic seismographs (as for instance, Benioff,
Grenet) the damping is achieved by electric resistances in the circuit con-
necting the seismometer with the galvanometer. By the shunt shown in
Figure 14 it is possible to reduce the magnification to a value which is
4.0 r - - - , - - - - - - , - - - - - , - - - - - - , - - - - , . - - - - - - , - - , 10000
logY v
5000
500
T sec
1.0'-----;!5:-----;-,10:;----;!;15:-------;2~0------::2'=5--~30:-l10
Fig. 15. Typical response curves for amplitude for some long-period seismographs. BB=Benioff N (Up).
WW=Wiechert N (Up). GG=Galitzin N (Ki). PP=Press-Ewing N (Um). V=dynamic magnification.
T=period. Cf. Table 2.
the load. As a consequence, shaking tables have been mostly applied for
testing short-period seismographs, especially those used in seismic prospect-
ing. However, more recent constructions have permitted testing with periods
of several minutes.
In other calibration methods, a device is used by which the seismo-
meter pendulum is set in motion of known amplitude and period, without
10 5
V
5.10 4
Up
10 4
510 3
any need to move the seismometer to any shaking table. There are several
different implementations of this method, for instance. by attaching to the
pendulum boom a small magnet, movable in a fixed coil. By connecting
an electric oscillator to the coil, the magnet and thus the pendulum are set
into vibrations which are recorded. In another calibration method. devel-
oped by the British seismologist WILLMORE, the seismometer coil itself is
used to put the seismometer into forced oscillations, where the seismometer
Pendulum Seismographs 51
---
A R B A. B = two piers firmly attached to bedrock at
a mutual distance of 20-50 m, R = rigid tube of
steel or fused quartz attached to B and extending
very nearly to A. A transducer for conversion of
mechanical vibration into electric current is
mounted in the air gap at A.
the ground at one end B, while the other end extends to another point A,
also attached to the ground. As such a system is hit by a seismic wave, a
relative motion occurs between the two points A and B. As a consequence,
the small air gap at A between R and A will vary, and this variation can
be measured. The usual construction is to provide this end, A, with a trans-
ducer, of the same type as in the variable-reluctance seismometer, and
then recording can be made in the usual way via a galvanometer. In some
installations, provisions have also been made for direct visual readings of
the air gap by means of a microscope; such readings can reveal longer-
period deformations in the earth. By using a quartz tube (R) it has been
possible to measure permanent deformations of 10- 9 , i.e. an extension (or
contraction) of 1 mm per 1000 km. For non-permanent deformations, the
sensitivity is about 10 times as high.
In actual constructions, the bar R is made of quartz because of its
low temperature coefficient. Earlier a steel bar was used, but with less
Other Seismograph Types 53
satisfactory results. The length of the bar R is usually of the order of 20-50
metres. In order to get a well-measurable relative displacement of the two
points A and B, the bar should exceed a certain minimum length; on the
other hand, the construction problems and the installation put an upper
limit on the suitable bar length. The strain seismographs, constructed by
BENIOFF, were made exclusively for measurements of the horizontal com-
ponent, but not the vertical component, of the relative displacements. On
the other hand, he has successively used a combination of two of his strain
seismographs, oriented at right angle)o each other.
BENIOFF installed his strain seismographs in some mountain tun-
nels in southern California. They are very sensitive to temperature varia-
tions (in spite of the quartz bar) and they have even been able to record
the daily expansion and contraction of the mountain where they are in-
stalled. Only deep underground installations have some prospect of lead-
ing to useful recordings. In connection with the International Geophysical
Year 1957-S, BENIOFF installed similar instruments also in South America
(Peru). Strain seismographs have been in use also ~:in Japan and in Ger-
many. They are still rather rare, which among other things is due to diffi-
culties in finding suitable localities for their installation.
The period response of a strain seismograph depends only upon the
galvanometer, as the seismometer has no free period of its own. By vary-
ing the galvanometer properties a possibility is offered to vary the seis-
mograph response within wide ranges. For instance, in combination with a
galvanometer of O.S-sec period, a response curve is obtained which corres-
ponds to the Wood-Anderson short-period torsion seismograph. As the
superiority of a strain seismograph compared with a pendulum seismograph
increases with the period, it is customary to connect the strain seismometer
to very long-period galvanometers, for instance, of 70 or ISO sec or even up
to S min (in Pasadena). It was with such a system with a ISO-sec galvano-
meter that BENIOFF was able to record a ground motion of 57 min period
from the great Kamchatka earthquake of November 4, 1952. This con-
stituted one of the strongest impulses to continued theoretical and ob-
servational studies of the free vibrations of the earth, generated by large
earthquakes. Such seismographs have also been used to record the daily
tides in the solid earth as well as to investigate the slow motions in the
solid earth, which give rise to earthquakes in seismically active regions.
54 Seismographs
480 stations and Japan about 155, there are only very few in large areas,
especially in the great oceans. In recent years, much effort has been devoted
to improvement of the station networks, partly by standardizing instruments,
partly by extending observations to new areas. The latter item includes
installation of unmanned stations in remote parts of the continents and of
stations on the sea bottom. Several experiments with the operation of ocean-
bottom seismographs have been made, particularly by the Russians and the
Americans. But still very much remains to be done, before the earth is
covered by an equally spaced station network.
Among instrumental improvements, I may mention the so-called
array stations (or multiple stations). They consist of a system of seismo-
meters arranged in some regular geometric pattern over an area extending
over a few kilometres to hundreds of kilometres. Such stations, built
mainly for the detection of nuclear tests, now exist in the USA, Canada,
Scotland, Sweden, Norway, India, Australia, and in a few other places.
In combination with recordings on magnetic tape and the use of big com-
puters, such stations offer a quicker and more complete analysis of the
recorded waves. However, special research has shown that even more tradi-
tional station networks can be treated as great array stations, as long as
the signals on the different stations show enough similarity that array
techniques can be applied. Tests of this kind on the Swedish network were
very promising, probably because of the similar ground conditions at the
different stations. Recordings of seismometers placed in deep bore holes
(so-called bore-hole seismographs) is another method which can be used
with advantage to avoid the disturbing background noise on the surface.
A suitably placed bore-hole seismometer could reach a higher signal sensi-
tivity than a considerably more expensive array station on the surface or
near to it.
As a consequence of more numerous and more sensitive stations,
the data flow to the world centres has increased enormously. This has led
to some reorganization of such centres, and since 1960 big computers are
regularly used for the calculation of earthquakes and other seismic events.
In Chapter 8 we shall describe in more detail how we are trying to
improve seismic records and their sensitivity. In Chapter 12 we shall learn
something about moon seismographs, which have been developed especially
for the observation of seismic phenomena on the moon.
56 Seismographs
1 Umea operated a Grenet-Coulomb Z seismograph in 1960-2, until the world-wide standardized equipment
of the U.S. Coast and Geodetic Survey was installed in 1962.
2 Uddeholm operated a Grenet-Coulomb Z seismograph in 1966-7.
3 The operation of Goteborg and Karlskrona was discontinued in 1968, and replaced by the more sensitive
station at Delary.
• Hedemora and Kungsor operated for one year (1969-70) together with Uppsala as a triangular array.
earthquakes, and then especially for the more long-period waves, as the
transverse waves and surface waves. For such recordings, the net does not
need to be particularly dense, and our three stations with such apparatus
(UppsaJa, Kiruna, Umea) fulfill the need very well. According to interna-
tional recommendations, fully equipped stations (like those we have at
Uppsala, Kiruna and Umea) should be placed at intervals not exceeding
1000 km. As is evident from the map in Figure 18, our net fulfills this
requirement very well. On the other hand, a denser net of Short-period
stations is needed, partly for a more complete recording of the short-
period components of P-waves from distant events, partly for recording of
near events, i.e. those which occur within Scandinavia and its nearest sur-
roundings. This requirement is well fulfilled together with our supple-
mentary stations (Skalstugan, Uddeholm, Delary).
The present network extends over more than 1300 km and exhibits
also a relatively even distribution over the country. Its extent and regularity
A Seismograph Network 59
permit good locations not only of near events but also of those at greater
distance. It should also be emphasized that our network ties in very well
with corresponding networks in the neighbouring countries. Seen from a
global viewpoint, there is no doubt that Sweden, like all the Nordic coun-
tries, is today well equipped with seismograph stations.
The combination of highly sensitive instrumentation and good bed-
rock has placed our network among the most sensitive ones in the world.
As an example of the sensitivity of the network, I quote here the number
of events recorded during the first six months of 1970:
Number of
Month events
January 377
February 371
March 458
April 804
May 538
June 524
60 Seismographs
The total amounts to 3072 events, i.e. on the average 16.8 events per day,
and this is approximately a normal figure. The majority of the events are
distant earthquakes but even a number of Scandinavian events (both ex-
plosions and earthquakes) is recorded.
When large earthquakes take place, these are generally followed by
a great number of aftershocks. Then, the number of recorded events is
considerably higher than just mentioned. One example is provided by
August, 1969, when after a Kurile Islands earthquake on August 11, we
recorded during the following day (i.e. for 24 hours) no less than 219
aftershocks. During the month of August, 1969, we were also able to
identify by means of our network 172 more aftershocks than reported from
anywhere else.
For one year, August 1969 to August 1970, the Seismological In-
stitute in Uppsala also operated a triangular array station. The three sta-
tions (Uppsala, Hedemora and Kungsor) formed an almost equilateral
triangle with a side of about 100 km length. The recording was made centrally
in Uppsala over the telephone lines.
In addition to the stations listed in Table 2, the Seismological In-
stitute at Uppsala has at different times operated about a dozen temporary
stations, partly in collaboration with the Research Institute of National
Defence, Stockholm. These stations have been located at places scattered
over the whole of Sweden and they have operated only for a few months
each. This has been done mainly for investigations of signal sensitivity
(Chapter 8).
61
Chapter 3
Seismic Waves
0)
p sv
, ..
b)
P SH L R SOURCE
- --I
Fig. 19. Sketch to demonstrate the propagation of the
direct waves in a slab: a) seen from one side, b) seen from
above. The arrows indicate the particle motions and the
waves have been arranged in order of their propagation
velocities.
shows schematically the particle motion and the propagation of the dif-
ferent waves.
We came across the different wave types already in Chapter 1. In
fact, it is the elasticity theory which can give complete information about
these waves, in other words, about the behaviour of materials upon com-
pression or dilatation (in the P-wave) and upon shear stresses (in the S-wave).
The conditions can be expressed in an exact mathematical form. In most
cases, however, simplifying assumptions are justified and acceptable in the
study of the conditions of the earth's interior. Thus, it is assumed that:
1. Relative displacements between adjacent particles are infinites-
imally small.
2. The material is perfectly elastic, i.e. the stress is a homogeneous
linear function of the strain and vice versa; usually, a generalized form of
HOOKE'S law is applied.
3. The material is isotropic, i.e. the elastic parameters are independent
of direction (the same in all directions).
4. External forces, such as gravity, friction, etc., can be neglected.
Main Types of Seismic Waves and Fundamental Laws 63
(1)
For the earth's crust and the upper part of the earth's mantle, we have
5
with good approximation that k = 3" J1 (corresponding to a POISSON
ratio = !' i.e. a ratio of ! between the lateral contraction "nd the
longitudinal extension of a cylinder). Substituting this into equation (1),
we find that for the upper parts of the earth Vp: Vs = y'3: 1 (cf. Chapter 1).
Now, we can get an explanation for the order in which the different
waves arrive (Fig. 19). The P-wave has always a higher velocity than the
S-wave according to equation (1) and therefore it always arrives before
S in the seismogram. The P-wave, which corresponds to the sound wave
through the earth, has the highest velocity among all seismic waves. No
reliable velocity difference has been observed between SH and SV. Nor
is any such difference to be expected in elastically isotropic media, while
with anisotropy a velocity difference may exist. The Love (L) wave is an
SH-wave, as we have seen, but it arrives later than S. The reason for this
is that during its propagation it is bound to the surface layers, where the
velocity is lower than in the interior parts of the earth, through which
the S-wave travels. Finally, the Rayleigh (R) wave arrives later than the
Love wave. This can also be proved theoretically. In the simplest case,
with Rayleigh waves propagating along the surface of a homogeneous me-
dium, it can be shown that their velocity is =O.92vs. Thus, they have a
lower velocity than the Love wave would have, and this result is not changed
if instead we assume a layered structure. Incidentally, Love waves require
a layered structure for their existence, whereas Rayleigh waves can also
exist on the surface of a homogeneous medium.
During their propagation through the earth's interior, the body
waves follow the same laws which hold for any other wave propagation,
for instance, in optics. The most important relation is SNELL'S refraction
law. In optics, it is generally written as n sin i=constant along a given
wave path, where n=the refractive index and i=the angle between the
incident wave direction and the normal to a discontinuity surface. This
angle is generally termed angle of incidence. In seismology, it is more
Main Types of Seismic Waves and Fundamental Laws 65
the wave velocities are changed. If the two media differ only in other pro-
perties, then the surface of separation has no influence on the wave pro-
pagation. In a study of Figure 20, we must remember that the P-wave,
being longitudinal, is able to propagate through any medium (whether
0) b) c) d)
*
p p SV P
P ~
p SV P
~ ~
P P P SV P P P
-* -+-
SV p p
SV ~
sv SV p sv SV P
SOLID LIQUID SOLID LIQUID
*- -+-
* -+-
SH
SH
SH SH SH SH
Fig. 20. Reflection and refraction of P, SV and SH at boundaries between different media: a) solid-solid,
b) liquid-liquid, c) solid-liquid, d) liquid-solid. In the figure, 'solid' and 'liquid' have been indicated only for
SV(middle figure). Cases where no wave paths have been drawn, cannot exist (incidence from below 'assumed
in all cases).
solid, liquid or gaseous), whereas the transverse wave (S) propagates only
through solid media. Moreover, we have to keep the particle motions in
mind (Fig. 19), from which it follows that, for example, an incident SH-
wave can only give rise to SH-waves (by reflection or refraction). Similarly,
an incident P-wave gives rise only to P- and SV-waves, but not to SH-
waves, and an incident SV-wave gives rise to P- and SV-waves, but not to
SH-waves. From equation (2), we also find that the angle of incidence i is
smaller for S than for P, because Vs is smaller than Vp. The atmosphere
can as a rule be treated as a vacuum, i.e. without any wave propagation.
Concerning the wave propagation through the earth's interior, there are
two facts which we have to take into consideration:
1. The earth is a sphere and all layers have the same curvature (as a
first approximation).
Body Waves from Distant Earthquakes 67
2. The properties of the earth vary with depth and in general the
wave velocities increase with depth.
Item I has a dominating significance, while item 2 is to be considered
at most as a correction to the results that item 1 leads to. This means that
we can get a very good apprehension of the wave propagation by studying
the conditions in a homogeneous sphere. This simplifies the discussion con-
siderably. In some modern wave propagation studies by famous theoretical
seismologists, it is also customary to assume a homogeneous sphere, as
its wave propagation approximates well to the real earth. We have to
understand this in such a way that the homogeneous earth provides for a
qualitative approximation, which will greatly assist us in understanding the
principles of the wave propagation in the real earth. On the other hand,
there may be quite large divergences quantitatively, i.e. when numerical
applications are made, between the homogeneous and the real earth.
0)
b)
u{pp
pp
Fig. 21. Principles of the wave propagation through a homo-
geneous sphere: a) surface focus, b) deeper focus. F = focus,
U = station.
paths. Figure 21 shows some of the waves reaching the station D. Besides
the direct waves P and S, there are those which have been reflected at the
earth's surface from underneath. The incident P-wave gives rise to a re-
flected P-wave, denoted PP, and to a reflected S-wave, denoted PS. Sim-
ilarly, by reflection against the earth's surface, the S-wave gives rise to a
reflected S-wave, denoted SS, and a reflected P-wave, denoted SP. This
illustrates how the notation for the seismic waves is built up. One symbol
(P or S in the cases just mentioned) is used for each part of the wave path.
In general, the S-wave leaving the focus F consists both of SH and SV.
However, if SV were missing in the direction from F to D, then there
would be no SP-wave at D. According to equation (2), the incidence angle
at the reflection point is bigger for P than for S. While PP and SS have
a common reflection point B midway between F and D, the reflection
points for PS and SP must be shifted from R The reflection point for
PS is located at C, i.e. closer to F than to D, and the reflection point for
SP is closer to D than to F. Applying equation (2) and assuming Vp: Vs =
= Y3, we find easily for the homogeneous sphere in Figure 21a and for FU =
= 180°, that CB=300, whereas for FD = 130° we find that CB=45°. We
also find that for PS, as distinct from PP, there is a certain minimum dis-
tance FD, below which PS cannot be obtained. For the homogeneous
sphere, this minimum distance is 110 0. For the real earth, the minimum
distance for PS is about 44°.
Then we can repeat the process, with two successive reflections
against the earth's surface. From an original incident P-wave, we thus get
PPP, PPS, PSP and PSS, and an original incident S-wave gives rise to
SSS, SSP, SPS and SPP. We have now three parts (from F to the first
reflection point, between the two reflection points, and from the second
reflection point to D), and as a consequence three letters are used in the
symbol for each wave. Obviously, we can continue this operation with
three reflections and form the corresponding symbols. However, it is quite
rare that more than two reflections from the earth's surface are clearly
readable in seismograms. In addition, we can consider wave propagation,
including reflections, along the greater arc between F and D in Figure 21.
After this, we shall study the wave propagation from a deeper focus
(Fig. 21b), instead of a surface focus, but still for a homogeneous sphere.
Then it can easily be shown that the law of reflection is fulfilled only at
Body Waves from Distant Earthquakes 69
90'
i,
PP
i2 i,
60
\
~~-- .. ~/
~
pP
30
I
0
~,=1.6' log~, "',= 71'
0.5 1.0 1.5 2.0
Fig. 22. Diagram demonstrating that the two reflection angles il and i2 are equal at two points on the smaller
arc FU. The curves have been constructed under the assumption that LI = 150' and Rj(R-h) = 1.1.
One of the reflection points (A in Fig. 21 b) is located near the focus and
obliquely above it, whereas the other (B) is nearer to the midpoint between
F and U. The first reflection point corresponds to the optical case, while
the other one is generally not observed in optics because of the limited
extent of mirrors. Waves leaving F in an upward direction (above the
horizontal plane through F) are denoted by lower-case letters: p for lon-
gitudinal waves and s for transverse. Besides the waves reflected at Band
already dealt with in the preceding paragraph, there are now also waves
reflected at A. An incident longitudinal wave at A gives rise to the reflected
waves pP and pS, and a transverse wave gives sS and sP. Just as in the
case of a surface focus, the reflection points for the transformed waves
(pS, sP) do not coincide with those of the non-transformed waves (pP, sS).
Also in the case of a deeper focus, it is possible to have several reflections
at the earth's surface, with the first reflection at A. Such waves are then
logically denoted by pPP, pPS, pSP, pSS, sSS, sSP, sPS, sPP. These
waves are frequently observed on seismograms. The deeper F is, the later,
clearly, is the arrival of pP in relation to the direct P-wave. For deeper
earthquakes, we get a greater number of phases than for a surface focus,
and in addition the recorded phases are usually sharper for the deeper
earthquakes.
70 Seismic Waves
PcP
PKP
~_-- P diffracted Fig. 23. Principles of the wave propagation in the case of two
A around the core concentric, homogeneous spheres.
Body Waves from Distant Earthquakes 71
30' PKP2
c
20
(VP)mantle:( VP)core = 1.2 and that the core radius is equal to half the earth's
radius and I have applied the wave-path equation (2). I want to emphasize
again that the intention in this and other model calc'llations concerning
the wave propagation is not to approximate the earth numerically, but
simply to bring the wave propagation back to simple principles, well known
from school physics. In the case mentioned (Fig. 24), i=300 corresponds
to the ray that is tangential to the earth's core. Then letting i gradually
decrease, we find that the distance to U first decreases from 187 ° down to
155°, where it 'turns' and increases again up to 180°, the latter distance
corresponding to i=Oo. The implication of this is that every point U sit-
uated more than 155° from F will receive not one, but two PKP-waves,
which have propagated along different paths. Exactly at 155 ° these two
waves coincide. This means a great concentration of energy at a distance
of 155 ° in our model (which can easily be seen by simple geometrical con-
siderations). We say that a caustic is formed on the surface of the sphere at
this distance. The two PKP-waves use to be distinguished by the notation
PKPI and PKP2 for the first and second arrivals, respectively.
The conditions in the real earth correspond to the phenomenon just
described for our model. But the properties of the real earth deviate from
those of the model chosen, and so does the location of the caustic. For
the real earth, we find it at a distance of about 144 °from the source. But
in principle the phenomenon is nothing other than what we know from
school optics as the minimum deviation for light passing through a prism.
Figure 25 illustrates the caustic phenomenon in a striking way. The existence
of a shadow zone and a caustic is a direct consequence of a low-velocity
layer at the inside of the outer-core boundary (see further Chapter 7).
Ki ni --- --
-..:.:.:..d::!;;Jl~
----
Sk ., ,I ,;,',
This is the most marked low-velocity layer in the earth, but corresponding
effects, though less pronounced, exist also from other less marked low-
velocity layers at other depths in the earth.
According to the simple geometrical wave propagation, there should
be a shadow in the range of 103°-144°. However, weaker signals are ob-
served also within this range, which can be explained by three diffe-
rent effects:
1. Diffraction of the P-wave around the core boundary. Such waves
have been observed up to a distance of about 165°. These waves are as a
rule of very long period (around 20-30 sec), and therefore they are best
recorded by long-period seismographs (Fig. 26). The corresponding dif-
within the shadow zone. The existence of the inner core has later found
full confirmation from numerous observations.
In fact, the boundary of the inner core is more complicated with
several layers, extending altogether over a few hundred kilometres in depth.
Therefore, it is more correct to talk about a transition zone between the
outer and the inner core than a simple discontinuity surface. Recent in-
vestigations suggest that some layering exists also at the outer-core boundary.
See further Chapter 7.
The inner core made it necessary to further supplement the wave
notation. A wave which has traversed the earth's inner core as a longitu-
dinal wave is denoted by I (this symbol refers to the part of the wave path
that falls within the inner core). As concurring evidence indicates that the
inner core, as distinct from the outer core, is solid, there should also be a
possibility of transmission of transverse waves through the inner core;
these phases are denoted by J. Thus we can form the following wave nota-
tions: PKIKP, PKIKS, SKIKS, SKIKP, PKJKP, PKJKS, SKJKS, SKJKP.
Naturally, the symbols I and J have always to be surrounded by K. Much
effort has been spent in searching for waves with J characteristics, but
still with no reliable results. This may be due to small amplitudes of the
corresponding waves. As mentioned above the boundary of the inner core
is rather a transition layer and not as sharp as the outer core boundary.
A consequence of a gradual transition to the inner core may be that the
K-wave is only insignificantly transformed into a J-wave, while nearly all
its energy passes into an I-wave. For similar reasons, there are not many
completely reliable observations of waves reflected from the outside of
the inner core (denoted PKiKP, as an example).
Without difficulty we can build further with the symbols already
introduced. For example, PKKP is a P-wave which has been reflected
once against the inside of the core boundary. PKPPKP or P' P' is a PKP-
wave which has been reflected once against the earth's surface. Also several
such reflections have been observed in seismograms, for instance P' P' P'.
As P' is strongest at a distance of 144°, we would also expect that the
strongest reflections are obtained at about this distance. The consequence
of this is that the strongest P' P'-waves will be observed at distances around
72°, and the strongest P'P'P' also around 72°. While P'P' is regularly
observed in seismic records, P' P' P' is more seldom seen. On records in
Body Waves from Near Earthquakes 75
Sweden it has been observed only a few times and then from earthquakes
in Mexico. P' P' arrives about 18-20 minutes after the PKP-phase. It
resembles a new P- or P KP-phase and earlier it was often misinterpreted
as such. However, p'P' in general differs from its parent PKP-phase,
partly by somewhat longer periods, partly by a less defined onset of the
wave (moreover, it is frequently doubled on the records).
The next step in developing the model, we have built up by means
of homogeneous spheres, is to introduce the velocity distribution that we
know exists in the interior of the real earth (Chapter 7). Figure 27 gives a
picture of the corresponding wave propagation. Since the wave velocities
generally increase with depth in the earth, the wave paths will be curved
and concave towards the free surface, according to equation (2). We have
seen that the homogeneous sphere models lead us remarkably far into
the understanding of the wave propagation in the earth, at least in principle.
The velocity distribution in the real earth entails at most modifications of a
quantitative nature in the picture we have arrived at.
However, in real seismograms there are a number of minor features
which are not incorporated in the picture so far presented. Such features have
nevertheless attracted much attention, especially in recent years by means
of array-station records, and they have given useful information about
factors of significance for the wave propagation. Particularly noteworthy are
the precursors or early arrivals, i.e. smaller-amplitude waves arriving a few
seconds and more ahead of the main wave. These have been observed for
PP in the distance range of 90°-110° and for PKP. Such precursors are
nowadays usually interpreted as due to scattering phenomena. Scattering
- earlier only looked upon as a disturbance - has certainly been raised
recently in importance as a wave-shaping factor.
For distances less than 10°, other complications enter into the wave pro-
pagation, this time depending upon the regional structure of the earth's crust.
Figure 28a shows a vertical section through a typical continental crust of
the earth. 00 is the earth's surface, CC is the Conrad discontinuity be-
tween the upper layer (granite) and the lower layer (basalt), and finally
MM is the MohoroviCic discontinuity which marks the base of the earth's
-.J
C1\
en
(l)
t;;.
3
n'
~
(€
V>
l~ttt\ E
,. S" 9" I:>" II " 12" l,lw I""
l ime --...,:..
Fig. 27. Vertical section through half the earth, showing the propagation of the longitudinal waves from a source in the left corner of'the figure. The
outer-core boundary is marked by a circular arc at a depth of 2900 km and the transition to the inner core is marked by a shaded zone at about 5000 km
depth. After B. GuTENBERG.
Body Waves from Near Earthquakes 77
crust. Below MM we have the earth's mantle with uItrabasic rocks. The
layer thicknesses given in Figure 28a correspond approximately to average
conditions on the continents.
From an earthquake source F in Figure 28a we have the following
direct waves propagating to the station U:
Pg, Sg=longitudinal and transverse waves, respectively, through the gran-
itic layer (sometimes the notation P, S is used for these waves);
P*, S* = longitudinal and transverse waves, respectively, which have fol-
lowed CC (sometimes also denoted by Pb, Sb);
Pn, Sn=longitudinal and transverse waves, respectively, which have fol-
lowed MM.
Evidently, it is necessary to further amplify the wave notation we
learnt in the preceding section, when we have to deal with distances less
than about 10°. Waves propagating along the boundary between two media
with different velocities travel with the higher of the two adjacent velocities.
In addition to the direct waves mentioned above, there are also reflected
waves, for instance, waves reflected from CC and MM.
For a 'typical crust under the deep ocean, we can make a correspond-
ing picture by excluding the granitic layer altogether and by diminishing
the thickness of the basaltic layer to about 5 km (Fig. 28b). However, for
a) b)
o' ____________~=_~
~
F ',water U
c: __+~____I____,L1~5~':. / - ~ F s:p*+-pasalt 7
// Pn
/
/
~_ _ _-L__~M//
M--
Pn
Fig. 28. Principles of the wave propagation through a continental crust (a) and an oceanic crust (b). F = source,
V = receiver, 00 = earth's surface, CC=Conrad discontinuity, MM = Mohorovicic discontinuity.
Fig. 29. Typical record of a relatively near earthquake. The shock occurred in Vastergotland, Sweden, at
58.4'N, 14.1°E at 2235 19 GMT on September 3,1968, and the record shown here was obtained at Uddeholm
at a distance of about 180 km. The first phase (Pgl) and the largest amplitudes (SgI) are easy to recognize
(cf. Table 3). There is I minute between successive time marks and time advances from left to right.
of 'long waves', i.e. surface waves, without making any distinction between
Love and Rayleigh waves.
Among the surface waves, the Love and Rayleigh types dominate
on the seismograms. Stoneley waves are not observed as a rule, as they are
restricted to internal discontinuity surfaces in the earth and the layers
adjacent to such surfaces. Love and Rayleigh waves follow the earth's
free surface and the layers just beneath it during their propagation. Their
properties are thus defined by the earth's crust and the upper part of the
mantle.
There are two main results which indicate that this part of the earth
is layered and not homogeneous:
1. Existence of Love waves. It can be shown theoretically that Love
waves cannot exist on the surface of a medium which is homogeneous.
They require at least one outer layer or a velocity which increases con-
tinuously with depth.
2. Velocity dispersion. For both Love and Rayleigh waves it is true
that the whole wave group does not arrive at the same time at a station, but
as a rule the longest waves travel with the highest velocity and thus arrive
first, followed by shorter and shorter waves. This phenomenon is called
velocity dispersion or simply dispersion.
The dispersion depends on the layering in the upper parts of the
earth, where velocities in general increase with depth. This is illustrated
in Figure 30. During their propagation, the surface waves extend to some
J\
0) v
- Direction of propagation
J'\
b) v Fig. 30. The influence of a
vertical velocity gradient on
the dispersion of surface
waves: a) smaller gradient-
smaller dispersion, b) larger
gradient-larger dispersion.
v=wave velocity, h=depth
below the earth's surface.
Surface Waves (Fundamental Mode) 81
depth in the earth, say, to a depth of the same order as the wave-length.
Rayleigh waves with periods of 60 sec would thus be noticeable to depths of
around 200 km, while those of 20 sec would affect depths only to about
70 km. If we assume the velocity to increase continuously with depth as
in Figure 30, it is obvious that the longer waves should propagate with a
higher velocity than the shorter waves. This is called normal dispersion (the
opposite case, when the short waves arrive first, is called inverse dispersion).
The more rapidly the velocity increases with depth the greater will the
difference in propagation velocity be for different periods, i.e. the greater
is the dispersion. In Figure 30 this is illustrated by two cases with different
velocity gradient and a correspondingly different dispersion. Instead of
assuming a continuous velocity variation with depth, we could have con-
sidered several superimposed layers each with a constant velocity, but in-
creasing downwards. This latter model is more common in applications,
as being more apt to the real earth. In practice, the procedure is the re-
verse: from a given dispersion curve, obtained from records, one should
calculate the corresponding structure. Here, as in any similar reversal of
a procedure, one has to consider carefully if the interpretation is unambigu-
ous or not. In most cases a given dispersion curve can be interpreted in
terms of more than one structure. If dispersion curves are available both
for Love and for Rayleigh waves, this limits to a certain extent the number
of alternative interpretations.
Especially earlier, the arrival time of the beginning of the long waves
(L) was frequently reported in seismological bulletins. Obviously, such in-
formation has no great significance, not even if one distinguishes between
Land R, since also the period has to be given. In addition, an analysis
of the whole wave train is necessary for a full understanding (this is, how-
ever, beyond the routine duties of a seismological bulletin).
A consequence of the dispersion phenomenon is that we have to
distinguish between two different velocities for the propagation of surface
waves: phase velocity and group velocity. By phase velocity or wave velocity
we mean the velocity with which a certain phase, e.g. a maximum, pro-
pagates. By group velocity, on the other hand, we mean the velocity with
which a whole wave group propagates. The two velocities are not equal
when there is dispersion, and this can be understood by realizing that a
certain phase (e.g. a maximum) can propagate in relation to the wave
6 Bath: Introduction to Seismology
82 Seismic Waves
group to which it belongs. A new wave appearing at the front of the wave
group can be followed through the wave group until it disappears at the
rear of the group; in such a case, the phase velocity is less than the group
velocity. Or conversely, a particular wave may be seen to move through
the wave group from its rear end until it disappears at the front of the
group; in the latter case, the phase velocity is greater than the group velocity.
This latter case holds for waves on a water surface, which incidentally
offer one of the best visual examples of surface waves. For body waves,
no clear indication of dispersion has been observed, and then the phase
velocity and the group velocity are equal.
For surface waves the amplitudes are largest at or near the earth's
surface and they decrease (roughly exponentially) with depth. The earth's
surface is an antinode for the vibration and at greater depth there are one
or more nodes, where the amplitude vanishes. Conversely, a given earth-
quake would generate the biggest surface waves if it is located near the
surface, whereas if located at or near a node it would give only insignificant
surface waves. This is also confirmed by observations in nature. This is
an immediate application of a well-known principle, which holds for every
kind of wave motion. In order to generate vibrations, the onset should
be made at an antinode and not at a node of the vibration in question.
In seismology this fact is generally termed RAYLEIGH'S principle. Most
earthquakes are located in the vicinity of the earth's surface (i.e. within
the upper 30 km) and therefore they give rise to clear surface waves; in
fact, the surface waves then have the largest amplitude on the whole record,
far exceeding the amplitude of the body waves. A typical case is shown
in Figure 31. However, with increasing depth of focus the surface waves
become smaller and smaller, and may be relatively insignificant compared
with the body waves. This fact provides a reliable means to distinguish
at first sight between shallow (or so-called normal) earthquakes and deep
earthquakes.
In seismological bulletins of older date, in a few exceptional cases
still, we find in connection with surface waves some further symbols, such
as C and F. By C (from Latin cauda) is meant the tail of surface waves
which follows their maximum displacements. This tail can continue for
several hours on the records of stronger earthquakes. It is not a phase
in the usual sense and, in fact, it is not covered by the usual theory of seis-
Surface Waves (Fundamental Mode) 83
~"--'------'-:s-:'--------""'-'"1"';"~~+'
-r---"--..............
-·-:~~~I'--
-' -·-I-·-·------t·'--...,'-i'-+T\Irj~~HLH~1r_! Fig. 3/. Records obtained at Uppsala by
. - .... j"'~-
, ' II
three long-period seismographs (Press-
Ewing) of an earthquake in Turkey (39.2'N,
I 28.4"E, focal depth 37 km) on March 25,
1969 (origin time 1321 32.4 GMT). Magni-
tude M = 6.1. The arrows indicate the di-
rection of ground motion, i. e. upwards on
the records corresponds to eastward, north-
ward and upward ground motion, respecti-
vely.
84 Seismic Waves
of the earth's crust and upper mantle, which are especially pronounced
in transition zones between continents and oceans. As some parts of the
surface-wave train will thus have propagated a longer path they will also
arrive later at any station. Alternative explanations consider the cauda
waves to be due to standing vibrations of crustal layers, generated by the
surface waves. By F (from Latinfinis) the time was indicated for the termi-
nation of any visible motion on the seismogram. Such an indication has
no real significance, as it depends very much on the sensitivity of the seis-
mograph used. We have to remember that the phenomenon at the earth-
quake source lasts only a very short time of the order of seconds. The
fact that an earthquake record at some distance from the source can last
for several hours depends exclusively on various wave propagation effects,
particularly dispersion, and has nothing to do with the duration of the
earthquake as such. For a given seismograph (of a given sensitivity) the
time interval F-L provides a certain measure of the earthquake magnitude
(Chapter 4), and has in some cases been used for such calculations.
Thanks to the installation of more long-period seismographs in recent
years a greater number of records of long-period surface waves, so-called
mantle waves, have been obtained. These can be both of Love-wave type
(usually denoted G after GUTENBERG) and of Rayleigh-wave type (R). They
frequently have periods of 8-10 minutes, which corresponds to wave-lengths
of more than 2000 km. Evidently, nearly the whole of the earth's mantle
must take part in these vibrations simultaneously. As the wave-lengths are
an appreciable fraction of the earth's radius, it is necessary to take the
earth's curvature into consideration in their study. A typical feature of
these mantle waves is that they can be observed repeatedly, as they are
passing around the earth. After the direct waves, which have travelled the
shortest distance from source to station, those mantle waves arrive that
have travelled along the greater arc from source to station. Some time
later waves arrive which have again gone the direct way, but which in addition
have travelled once around the earth, and so on. How the mantle waves
are denoted is evident from Figure 32. An exceptional case with very clear
mantle waves was offered by a strong earthquake in Chile in May 1960
(Fig. 33). At Uppsala mantle waves were recorded up to 60 hours after
the earthquake. These must have been able to encircle the earth about 20
times before they were too weakened to be recorded. G20 or R20 have
Surface Waves (Fundamental Mode) 85
travelled a distance which is about equal to the distance from the earth to
the moon. In many other cases observations have been made up to G8
or R8 and over. Occasionally, mantle waves are also reported in seismological
bulletins of older date, but naturally only for the strongest earthquakes.
In these bulletins they were usually denoted by W (from German Wieder-
kehr-Wellen); W2 corresponds to R2 and W3 corresponds to R3.
Closely related to the mantle waves are the free vibrations of the
earth, which are generated by stronger earthquakes and which have also
been recorded a number of times in the last decades (e.g. Kamchatka
in 1952, when BENIOFF suspected a wave with a 57 minute period, and
notably Chile in 1960). In the interval between these two earthquakes,
the theory had been worked out, especially by PEKERIS and his group in
Israel, and the Chilean earthquake in 1960 gave a brilliant confirmation
of the theories developed. Another early observation refers to torsional
oscillations from an Alaskan earthquake in 1958 (BATH, 1958). Since 1960,
free vibrations have been recorded and studied from several earthquakes,
especially Alaska on March 28, 1964, and the Aleutian Islands on February
4, 1965. A useful summary has been published by DERR (1969). The free
vibrations are of different kinds: spheroidal (including radial vibrations)
and torsional (without radial component). See Figure 34. Like all free vibra-
tions these are standing waves. The spheroidal waves arise through mutual
interference of propagating Rayleigh waves, and the torsional vibrations
00
--:-"- . - ~ 0..
, Nr.~T'J ---------
:::::.:::::-:-:::~,-'--;-:-:-;"~- i ' :: . : : ; l ,
.... - - [ , " - : ' , , - - - _'--'" . j 'j ; ., p', .. I I .' : i 't ' 1 ... : •#' ------....,I-~.\\~ll't:/\:...... ""',, .'' .. (",:, ~.i.#,l-'·t Ip ;\.,\:4,<J..':~
- , - - -'- ' __'_'-r·~,""~\.~~I\"""V~f~\. YtM4... ."..'y': "/. )/'h :.~(~\:.'.~~t\·\', \~ ,1.1 -\, .; .. ' -/'\14, \. (I}
.. ~ ~ ~ . '" ,; ~. : ', · ', .~ "I ;: " ", (\ A f , ; •.':',:' ,? . . :~"It· ,':1:' ~!)j~ I" • • ,\ '~ 7'~ J:'I 't:- ~ .t:.. <I>
- ---
.g~~~~~ ~-----, , >"I
~ in '
_~'_01 '~~
§,~~~,.:. . . . .,~...,~,~~~
~ .~'~'~~.~'~.'. ~"~.'~"~~,\~':~'I'.' ~.~"':~: :'§;-~ .'~'~'' ~' ' ~':'~'~ 7' I~i~ ' ~·: ~'~) ' '~~'i·:~'>.iJ§'.~;t~"~~~~')~N~~~'~'.~.~,!~.~;:~.,~.~~,.........
§~
~~§§§§§~ . .- - ..'_"" . ,
. 3
n'
---' - -- ~
'-'- '---'--~...... :: .....'1':".', Po>
iiien
~. :Kft
;t " -1'
I. , . ....----~. -:-~::::::: :C
------.,-~.,'" ----... '-..:..-,.,.---,~~.,.. ../,.,-"'-~. --~~~--,-'"---.------ ,
~'-:-~-_~ . . .,-7
. "- . ./~ ~____ _ .- --........ ~ _. ......... ~ -~._"'.._-----.,....-
~'........., ' -----~" " '----.:..-..-."""""-..,-~
. . . - -.
~~-.--
....- [. ". - ..--~----.
-'---"- "--- -
__ ---..----....._~~
- ' - - --.-----:::::--:'>', -,.-:V. ..·~ .."_:...\Iy., ....;~••,. ·,' .......-~"_~--- ------
~.,---r----'~·--:-:--.....--------~---~-,,-.---------_·-------~·~
. "'- . .- . .
::~~~=~~s~~=-~~
=-==--~~~-~~~ .:-=:===--:===::==-==-~~ .:;-~-- --:::::~~~
----.- -, :~::-:::==:==='===::=::::::=--'---'--'------'--,. '- :-'-:-" j *'
- ' '_ ' .-.-.:...-,- '- ' ---- ~ -w-'" ...J
~-.-.-.----.--.---'~-,------ ---.--...-'- -=-"- ' '-.
~ '4
--------.- '- -'-----.--"-'----:-----"---,-~=--==== .-~-="=====-==: .~
:.:.:=~:~-=:====~:-':::===~.~::=:::~~.~ -~_.~~::-~ -:"=~=:=:::~::3--'-r~==:;~ .
Fig. 33, The Chilean earthquakes of May 22, 1960, according to records of Press-Ewing seismographs at Uppsala. The upper picture comprises May
22-23 (with the Agadir earthquake of February 29, 1960, inserted for comparison in the upper right·hand corner), the lower picture shows the conti-
nuation for May 23-24, 1960, with clear mantle surface waves, There is 1 minute between successive time marks and time advances from left to right
(1 hour for each line. proceeding from top to bottom).
Surface Waves (Fundamental Mode) 87
t
Fig. 34. Lowest modes of the free
torsional and spheroidal vibrations
of the earth. Mode notation is
explained in Table 4c). Nodal sur-
faces are marked by dashed lines,
and arrows indicate the direction
of motion. U =upward motion,
o = downward motion. After BAR-
BER (1966), modified.
arise from Love waves in the same way. Not only fundamental vibrations
but also those of higher modes ('overtones') have been investigated theoreti-
cally and observed in records. The longest theoretically calculated periods
amount to 44 min for the torsional vibrations and 56 min for the spheroidal
vibrations. Spheroidal vibrations of the earth's core have been calculated
to have periods up to about 100 min. The observations of the earth's
free vibrations have not caused any essential change of the picture of
the earth's interior which had already been arrived at by other means.
Observed and theoretically calculated periods agree within 1 %. Comparisons
have been made both with BULLEN'S and with GUTENBERG'S earth models
(Chapter 7), and a somewhat better agreement was found with the latter
model (including a low-velocity layer in the upper mantle). Observations
of torsional oscillations have only been of limited help in discriminating
88 Seismic Waves
From physics we know that a vibrating string, fixed at both its end points
A and B (Fig. 35a), is able to perform a fundamental vibration, in which
the wave-length is equal to double the length of the string. But, in addition,
the string can at the same time perform a number of vibrations of higher
modes (overtones). Mathematically we have the following simple rela-
tion between the length I of the string and the wave-length ).:
A.
I=n- (3)
2
where n = I for the fundamental vibration, n = 2 for the first overtone, n = 3
for the second overtone, etc.
Similar conditions prevail for the surface waves in the earth, and
we can illustrate this simply with a Love wave or SH-wave propagating
in a layer (Fig. 35b). In order that a wave should be able to propagate to a
greater distance constructive interference is required. This means, for in-
stance, that the down-going wave AB should be in phase with CD which
propagates in the same direction. Just as in optics we can express this
Surface Waves (Higher Modes) and Channel Waves 89
Q)
n =1
A
~ --- ------
-----
----
.
~----
---- ------ . a
n =3
----
----
/
',:::,
~
7'VV'
b)
c
(4)
Fig. 36. An earthquake in Sinkiang, China, on November 13, 1965, recorded by a long-period vertical-
component Press· Ewing seismograph at Kiruna. The distance is 41.5 0 and the magnitude M is 7.0. PP
signifies the P-wave that has been reflected once from the earth's surface about midway between the epicentre
and the station; SS is the corresponding reflected S-wave. The higher-mode Rayleigh wav!s are exceptionally
pronounced.
Surface Waves (Higher Modes) and Channel Waves 91
the higher modes have shorter periods than the fundamental mode and,
moreover, they have higher velocities. Therefore, they arrive ahead of L
and R, but like the fundamental modes they exhibit dispersion with a wave
train which lasts for several minutes. See Figure 36.
The so-called channel waves are another but related type of waves.
In Figure 37 we have depicted a layer with decreased velocity (a low-
velocity layer). The refraction law, equation (2), requires that every wave
path which is not too steep towards the channel axis should be refracted
back to the channel, and this holds both above and below the channel
axis. Obviously, such a channel preserves the energy once fed into it in
a very efficient way, and in fact such channels are the most capable guides
known in nature. Such low-velocity layers are well-known from the atmo-
sphere and the seas (for sound waves). An explosion in the ocean can emit
part of its energy into such a channel in the sea. A longitudinal wave travels
along the channel until it strikes a coast, where it is transformed into the
usual land-propagated seismic waves. Earthquakes near the sea bottom
are also able to generate longitudinal waves through the ocean, but in this
case they generally propagate by mUltiple reflections between the bottom
and the surface. Longitudinal waves through the ocean are termed T-waves
(T as in tertiary, corresponding to P=primary and S=secondary). As
their velocity during the water propagation is only about 1.5 km/sec, they
arrive several minutes after the corresponding P-wave which has passed
through the earth's interior. See Figure 38.
In the earth's interior there are similar channels or low-velocity
layers. The most significant is situated at the inside of the outer core bound-
ary at a depth of 2900 km but concerns only the P-wave. However, there
are no sources (earthquakes) in this layer which otherwise would be able
to feed much energy into it. But waves which have been multiply reflected
92 Seismic Waves
Fig. 38. Typical case of a so-called T-phase, recorded by Benioff Z in Bermuda from a North Atlantic
earthquake (lO.6°N, 43.7°W, depth 25 km, magnitude M=6.8) at a distance of 3231 km, on March 17, 1962.
The travel time of Tis 35 min 21 sec, yielding a velocity of 1.52 kmjsec through water. The upper trace shows
P, the lower T.
in this layer are recorded, e.g. PKKP, PKKKP, etc. These may be con-
sidered as a kind of leaking channel wave.
Among low-velocity layers nearer to the surface, for which it is
maintained that channel waves have been observed, the following are the
most important:
1. A layer in the upper mantIe of the earth. This layer has been
generally accepted since around 1960, but was originally proposed as
early as 1926 by GUTENBERG. Pa and Sa (a stands for asthenosphere,
a name for the upper mantle) are longitudinal and transverse waves, re-
spectively, which have been explained as channel waves in this layer. As a
rule they have periods of 10 sec and 20 sec, respectively, and they are
often recorded by long-period instruments. Alternatively, these waves have
been interpreted as higher-mode surface waves.
2. Layers in the earth's crust, both in the granitic and the basaltic
media. Such low-velocity layers, suggested during the 1950's, especially by
GUTENBERG, did not find general acceptance. However, waves are ob-
served which have been characterized as the corresponding channel waves
(Li, LgJ, Lg2, see Table 3). Crustal low-velocity layers have experienced a
certain revival in the last decade, as several investigators deem them neces-
sary to explain their observations.
Most of the channel waves mentioned here have been found to bear
relation both to the higher modes of surface waves and to the waves found
on records of near earthquakes (see above, this chapter). Thus, it has been
possible to explain most or all of the channel waves in terms of higher
Surface Waves (Higher Modes) and Channel Waves 93
Pn Pn 7.8-8.2
Sn Sn 4.6-4.7
p* or Pb 6.6
S* or Sb Li 3.7-3.8
PgJ IIg 6.25
SgJ LgJ 3.54
Pg2 or Pg or P 5.7
Sg2 or Sg or S Lg2 3.37
Rg 2 Rg 2 3.02
1 These may be taken as characteristic values, but minor regional variations occur.
• Rg is a short-period surface wave of Rayleigh type.
modes which eliminates the need for a low-velocity layer for their pro-
pagation. On the other hand, it is clear that the velocities of the channel
waves exhibit such remarkably good agreement with the velocities of waves
from near earthquakes that a pure coincidence is excluded. However, in
both of these comparisons there still remains much to be investigated. For
this purpose, it would be particularly useful to have records from densely
spaced stations, from short distances out to great distances from seismic
events, preferably along a few profiles. Only in such a way would it be
possible to investigate details of the wave propagation, possible mode con-
versions, etc.
The wave symbols we have introduced in this chapter are summarized
in Table 4. The principles for notation of phases are generally accepted
and used by all seismologists. The notation has the advantage of giving
immediate information of the way in which a wave has propagated through
the interior of the earth. It is also easy to check that the system is fully
unambiguous, i.e. a given notation cannot stand for more than one wave.
In defining notation for new waves discovered in records or theoretically,
it is also of importance to conform to the system already developed.
94 Seismic Waves
In this table, P can everywhere be exchanged for S and I for J. P' and pI/ are used as
briefer notations for PKP and PKIKP, respectively: P'=PKP, P'P'=PKPPKP,
PI/=PKIKP. PKHKP=a wave with its deepest penetration in the transition zone
between the outer and the inner core.
c) Surface waves:
d) Special symbols:
'Channel waves': Pa, Sa, PL, Li, Lgl, Lg2, Ilg, T, etc.
Several of these can also be grouped under c).
Travel-Time Diagrams 95
One of the most important problems in seismology since the turn of the
century has been the determination of travel times (also called transit
times) with the highest possible accuracy. The corresponding tables or
graphs give the travel times of all seismic waves in dependence on epi-
central distance and focal depth. Obviously, this problem can be solved
only by successive approximations, at least when uncontrolled sources (earth-
quakes) are used. We start from an assumed source (with a certain latitude
and longitude) and an assumed time for the occurrence of the earthquake,
both obtained, for instance, by direct observations or otherwise. Then the
onset times from seismograph records at different distances permit the
construction of a preliminary travel-time curve which gives the dependence
of the travel time on distance. This curve can then be used for a more
accurate determination of the location of the source and the origin time
which in turn gives another, more accurate travel-time curve, and so on,
until the curve does not change. In all determinations of such diagrams it
is assumed that all seismic waves are generated by one and the same motion
at the source, i.e. all waves are assumed to have the same origin time.
Also, the source is assumed to be a point source.
A number of travel-time tables have been constructed. Among those
best known and most used are those of GUTENBERG-RICHTER and of
JEFFREYS-BuLLEN. The latter were first published in 1940, of which those
for a surface focus are shown in graphical form in Figure 39. The ordinate
t is the travel time and the abscissa A is the geocentric angular distance
between the epicentre and the station. As the arrival time of the P-wave
is the one which can be determined with the highest accuracy the same
holds also for the corresponding travel times. Obviously, the travel times
of different waves are not independent of each other. For instance, the
travel times of PP and of PPP are related to those of P by the following
relations:
tpp(A) = 2t p(A j2)}
(5)
tppp(A) = 3tp(Aj3)
provided that both focus and reflection points are located at the earth's
surface. Such relations are used to calculate the travel times of PP and
PPP from empirically determined times for P. Comparisons with seis-
'",
96
~m,,, .. tul
Fig. 39. Travel-time diagram for a surface focus according to H. JEFFREYS and K. E. BULLEN.
Travel-Time Diagrams 97
mograms have made it possible to identify PP and PPP. But the observed
times of P P and P P P may show deviations from the calculated ones, because
of deviations from the assumptions made in the calculations (focus and
reflection points at the earth's surface). That foci are deeper is of course
a general occurrence, but it has also been found, especially for the distance
range around 90°-110°, that PP exhibits early arrivals, due to reflections
at discontinuities within the earth's mantle. These early arrivals are generally
weaker than the surface reflections which follow later in the same record.
Even though the early PP has also got alternative explanations (Section 3.2),
there seems to be no doubt that deeper reflections do occur. In an analogous
way the travel time of PS can be deduced from the times of P and S:
Similar relations can be formulated also for other composite waves. In the
Jeffreys-Bullen tables, empirical times are used for P, S, PcP, ScS, PKP
and SKS as a base and all other times are calculated from these.
Determination of accurate travel times has naturally involved an
enormous amount of work, including a comprehensive statistical treatment.
On the other hand, the travel times have no doubt given the most important
information about the earth's interior. Some conclusions may be men-
tioned here. We see from Figure 39 that the travel-time curves of P, S
and all other body waves are curved in such a way that waves from greater
distances have travelled during part of their path with a greater velocity
than exists near the surface. The curvatures are such that they can only
be explained by a velocity which increases with depth in the earth. See
further Chapter 7. In comparison, the travel-time curves for LQ and LR
in Figure 39 are rectilinear. This means that these waves have travelled
along some layer of constant velocity, in this case the surface layers.
In general, the travel time of any body wave is a function both of
the epicentral distance and of the focal depth. By means of considerations
of the wave propagation in homogeneous spheres, made earlier in this
chapter, it is relatively easy to demonstrate that the travel-time difference
S-P depends much more on distance than on depth. Similarly, it can be
shown that the difference pP-P depends much more on focal depth than
7 BAth: Introduction to Seismology
98 Seismic Waves
on distance. These conditions prevail also in the real earth. See data in
Table 5.
A remarkable result is that the travel times for body waves for a
given distance and depth are very nearly the same, independently of the
position of the source and the station, after eliminating the influence of
the earth's ellipticity. This would mean that the earth has a symmetrical
structure (inside the crust). In fact, this is only an approximation, but it
is so good that even today the same travel-time tables are used as were
constructed around 1940, without making any distinction between
different parts of the earth. Howe\ler, improvements have been made more
Table 5. Some travel-time data.
a) Travel times of P- and S-waves and the time differences S-P for different distances
(..1) and a surface focus. Calculated from JEFFREYS-BULLEN tables.
.d'
P S s-p
min sec min sec min sec
10 2 28 22
4 1 54
20 4 37 17
8 3 40
30 6 13 11 10 4 57
40 7 38 13 45 6 07
50 8 58 16 09 7 11
60 10 11 18 23 8 12
70 11 15 20 26 9 11
80 12 13 22 17 10 04
90 13 03 23 55 10 52
100 13 48 25 20 11 32
b) Travel-time differences S-P (min sec) for different distances (..1) and focal depths
(h). Calculated from GUTENBERG-RICHTER tables .
20 3 34 3 24 3 12 3 07
30 4 51 4 37 4 26 4 16
40 5 57 5 41 5 27 5 17
50 6 59 6 44 6 31 6 19
60 8 02 7 44 7 30 7 18
70 8 54 8 37 8 22 8 09
80 9 48 9 31 9 18 9 05
90 10 42 10 25 10 10 9 56
100 11 26 11 09 10 51 10 37
Microseisms 99
20 0 16
30 0 19 0 57
40 0 23 1 00 1 29
50 0 25 1 05 1 36 2 03
60 0 25 1 07 1 41 2 08
70 0 26 1 10 1 47 2 15
80 0 26 1 12 1 51 2 22
90 0 27 1 12 1 52 2 27
100 0 27 1 13 1 52 2 28
3.7 Microseisms
a) 1 min 1 min
>---t
1min 1min
c) t----t d) ~
Fig . 40. Microseisms within different period ranges: a) short-period, T--<2 sec (Karlskrona Z, February 7,
1965), b) medium·period, T-;;,6 sec (UmeA N, April 5, 1969), c) medium-period, T=9-lOsec (UrneA Z,
March II, 1969), d) long-period, T= 17-20 sec (UmeA N, April 24, 1965).
Chapter 4
4.1 Parameters
2 3
Fig. 41. Determination of the position of the epicentre (E) by means of the hyperbola method (a) and by
means of the circle method (b). I •• I •• I. are the arrival times for the P-wave (or any other given wa,e) at the
three stations I. 2. 3. respectively.
1 2 3 4
J/))III/iJ/II!!II'l''1I1l! nh',n,J),; )lln,n/llm/III}I! 'I)}} ) II h.' 1'}.i,Vhf, Jlm/nfJlI";
\ I
\ I
\ I
\ I
\ ,, /
/
depth and origin time. In this schematic two-dimensional case, where the
epicentre is defined by only one coordinate, in fact three P-observations
would suffice.
The arrival times of P are usually the primary data with which we
have to work. But if, in addition, information is available on azimuth ob-
tained from the P-wave or on arrival times for other waves (for instance S),
then the number of stations can be diminished correspondingly. The follow-
ing two cases give some typical examples:
1. Two stations: station 1 gives P-time and P-azimuth, station 2
gives P-time.
2. One station: P-time, S-time and P-azimuth given (see further below,
this chapter).
In both the cases mentioned a unique solution is obtained under the
assumptions of a surface focus.
Even though the seismological literature, especially that of older date,
offers numerous examples of such geometrical games, these methods are
generally of less significance today. The common method today is to base
the determinations on as many different stations as possible. Local de-
viations exist from assumed travel-time tables depending, for instance, on
local structures, and therefore a least-square calculation technique is applied.
Then, the error limits of the results obtained can be indicated. It is cus-
108 Source Parameters and Their Determination
0) 01
09
010
07
o
5
b)
I
lfs10
E-E o
30 08 \
b1 6E 01
20
07
10
60
Az(E)
0 Fig. 43. Principles of deter-
180 6 270 360· mina tion of the epicentre by
6E
/5/
0
residual calculation using ob-
-10 servations at a number of sta-
tions. 1, 2, ... , lO=stations,
E = assumed epicentre, Eo =
-20
actual epicentre.
Coordinates and Origin Time 109
AUG GMT LAT LC~G REGION AND COIIIIEIITS DEPTH CGS MAG ~O N
H M S KM M8 MS
2 07 C7 02.5 J9.5 N 111.0 W UTAH 6 t.9 9
3 10 02 17.3 33.0 N 116.2 W SOUTHERN CALIFORIHA 1 3.8 C.I> 7
MAG. 3.2 "A5I.
10
5
15
06
23
21
13
07
25.1·
J6.4
04.7
37.6
J9.1
39.5
N
N
N
111.10 .
112.3 II UTA.H
UTAH
111.0 W UTAH
33R
15
8
10.0
1.6
0.8
t.8 8
b
8
5 23 10 28.7 39.5 N 110.9 II UTAH 5 K 1.0 q
11 lZ. 07 56.9· 1.7 N 126.3 E MOLUCCA PASSAGE 33 R I.Z I.
12 11 107 33.~* 1.7 N 126.7 E MULUCCA PASSAGE 33R 1.1 7
lZ lJ 58 33.Z. 29.6 N 141.1 E SOUTH OF HONSHU. JAPAN 108 4.10 0.4 9
13 11 ~9 16.1- 60.3 N 153.7 W SOUTHERN ALASKA 127 10.3 C.I> 14
14 10 13 1t6.1- 1.7 N 126.3 E MULUCtA PASSAGE 33R 4.7 C.7 10
II> 15 ~6 Z5.8· 21.Z S 179.3 FIJI ISLANDS REGION 667 10.2 1.0 15
II> ZI 33 106.7. 41>.10 N 14.2 YUGOSLAVIA H 10.': 1.0 7
FEL T AT RADUVLJlCA AND GORJE ANU
TRLlt.
H 13 18 101.]· 1.8 N 126.10 k MOLUtCA PASSAGE 33 R 1.5 10
21
22
o. u8
01 33
39.5· 23.7 S
12.1 Z9.8 S
UO.O E SOUTH UF FIJI ISLANDS
116.9 W KERHADEC ISLANDS
5~5
39
4.5
'.10
I.e
1.'
•
1~
21
24
11 01
12 21
210.8" 12.9 N
Z8.1 51>.2 S
1105.2 E SUUTo. OF MARIANA ISLANUS
1101.5 10 SUUTH PACIFIC CORDillERA
.5
33M
4.5
5.~
1.3
1.0 l'
•
210 I ' ZI> 01.1t- 30.0 N 95.1 E TI~ET 5_ ".1> 0.5 8
Z5 00 11 33.Z· I.Z N 1.0.1 E M"LUttA PASSAGE .l 5.3 1.1 11
26 05 59 09.1- Z6.7 " 55.0 E SOUTHERN IRAN
JJ "
4.6 1.2 ~
..
26 .,7 01 l i d "0.3 N 11ol.6 E OFF EAST COAST O~ HuNSHU, JAPAN Z. 4." 1.1 11
2. I" 35 08.5 0.2 S 121.8 E NORTHERN tELE6ES 295 ".5 I.Z Z.
"
26 17 210 35.C 17.6 S 69.6 10 PERU-~OLIYIA ~DRUER REG I 0"'" 170 ".1 1.0 1
26 ZO z.r. 28,)· 21.3 S 179.5 W FIJI ISUNOS REGIUN 1>71 ~.4 11
21 u6 ~3 "1.4. 19.2 S 111.1 W FIJI I SLANDS REG ION 6Jl '. C ~.~ I"
27 Zl 55 "8.3 1.0 N 120.1 E· NORTHERN CkHeES 31 R 5.C 1.~ Z1
28 OZ 55 30.~ 45.4 N 13 •• 7 E NEAR f. COAST .JF EASTERN RUSSIA 2108 ".Z ~.I 11
le 1~ os 31.5 1~.1 S 161,J E NE. HEBRIDES ISLANDS 123 4 •• 1.1 21
FELT AT LAMAP.
19 110 Z6 22.9 5.4 ~ 1"5.5 EAST NEW GUiNEA REGION 8b ~. ~ 1.1 12
29 Id Z1 08.9. 15.3 N 12Z.6 PHILIPPINE ISLANDS REGION 33M ~. > I., 5
19 ZI 15 "10.1 •• 9 N 13.0 W NUIUHERN toLUMBI A 151 '.1> 1.1 15
)0 O. 18 30."· I."
N 126.3 E MoLUtCA PASSAGE so 5." 1. Z '1
"
30 21 11 ZO •• 310.9 59.5 E IKAN 1) R • .3 I>
PKu8AaLE fliRESHOtK OF IKANJAN
~UAKE Of AUG. ] 1, HYPOcet.lYI::M
POuRLY OHERMINEO.
31 IJ H 37.4 H.~ N 59.0 E IRAN 11 •• 0 1.1 1.5 51>
MOR~ THAN 11000 K ILLEU ANO 6000
INJUREU. KUHAK .0ESTRoYELl. ~IGh SEE CrtMENTc AT roTTm
LlE~TRUCTJoN IN SURROUNDING AREA. PAr,E 2.
FEL T THROUGriOUT KHURAS'SAN
PROYIN~E. MAG. 7-1 11' IPASI.
7.7 IdRKI. 1 liZ I Gall.
31 11 H 32.9 H.~ N S9.Z IRAN 24 S.~
".~ H
11 19 06 3~.1 ~6ol N 115.6 EAS T OF LAKE BAIKAL .~ R ".1> 0.8 19
SEP
.
".
1 00 19 5".8 1.6 N Il6.3 .E "OLUttA PASSAGE l3 k 4.9 C.9 IB
1 01 50.5· 0.9 S 210.6 W CENTRAL MID-ULANT IC RIUGE 13 '.3 O.S 7
1
I
03
05
42 06 •• • 0.9
18 01.1. 0.9
S
S
210.7
24.9
W
•
CENTRAL "iD-ATLANTIC RIOGE
CENTRAL MID-ATLANT IC RIDGE
H R
31 R "."
10.10
~.8
1.v
6
9
1
I
01
01
03 39.9·
Z1 30.Z H.C
101 S
N
24.9
58.2
W
E
CENTRAL HID-ATLANT I t RIDGE
IRAN
33
15
• "."
5. C; b. J
0.0
1.0
9
01
MCRE THAN ZOOO KILLED ANO
EXTEIliS I YE PROPERTY UAHAGE AT
FERDOWS
HAG. 1>.2 'PAS" 1.0 I8RK), 6 1/''-
6 liZ IGoLl.
Fig. 44. Sample page of preliminary epicentre determinations from the U.S. Coast and Geodetic Survey,
now the U.S. Geological Survey. MAG = magnitude, SD=standard deviation of the origin time, sec,
N = number of stations used in the solution.
112 Source Parameters and Their Determination
4
p Z R
N
h
-
~
W AE d>
E
tN
--
./
./
./
...::::::'-- Of = Qzimuth
~
S A
tan a =A:
Fig. 45. Sketch showing how from ai~en records (E, N, Z) of the P-wave it is possible to determine the
direction (azimuth) to the epicentre. The corresponding Rayleigh waves are shown to the right, where one
should note the 90° phase shift between Z and the horizontal components. Compare Figure 19.
atively good accuracy, it has proved more difficult to obtain the focal
depth with the necessary precision. As indicated in the preceding section,
the accuracy of the focal depth, as measured in kilometres, compares
favourably with the accuracy in kilometres of the epicentre location. How-
ever, the requirements on focal-depth accuracy are more stringent. This
concerns especially shallower depths, i.e. for events in the upper 100 km
of the earth. The requirements on accurate depths stem from both prac-
tical and theoretical aspects of seismology, which can be summarized in a
few points:
I. Nuclear explosion detection. Events for which the depth exceeds
a few kilometres, say 5-10 km, are very likely natural (earthquakes), whereas
shallower events may be artificial. See further Chapter I I.
2. Tsunami warning. Tsunamis or seismic sea waves may arise be-
cause of earthquakes near an ocean bottom (Chapter 1). It seems probable
that only very shallow earthquakes give rise to tsunamis. Therefore, im-
proved precision of the focal-depth determination would be of immense
importance to warning services.
3. Pure seismology. In many seismological studies, a higher accuracy
of focal depths is required, especially for the upper 100 km of the earth.
This concerns every investigation where focal depth enters as an independent
variable, for example in investigations of spectra as a function of focal
depth or of energy content in higher-mode surface waves as a function of
focal depth, etc.
At the same time as the requirements are especially stringent for
focal depths less than 100 km, the difficulties for depth determinations
are greater in this range than for deeper events. Table 5 gives information
on the time differences pP-P which correspond to different depths from
100 to 700 km. For depths less than 100 km the problems of accurate depth
determination increase, partly because of the fact that pP follows closely
after P and may be difficult to read, partly because of greater influence
of local source conditions on the observed time difference pP-P.
The world centres mentioned above often do not consider it
possible to guarantee a higher precision than ± 25 km in the calculated
depths (even if these are given to 1 km). However, a higher accuracy can
be achieved by the simultaneous inspection of the records from a whole
network of stations. By direct visual comparison, it is possible to make
8 Bilth: Introduction to Seismology
114 Source Parameters and Their Determination
normal depth (h constant) and finally by stating that the function f for
surface waves is proportional to log ,1, equation (1) is simplified to the
following form:
(2)
where C1 and C2 are constants. In most formulas of this type which have
been developed the amplitude a refers to the horizontal component of
Rayleigh surface waves. Some similar formulas for the Rayleigh-wave ver-
tical components have also been developed.
3. m is magnitude based on body waves and defined by the same
equation (1). In 1945, GUTENBERG extended magnitude determinations also
to body waves (P, P P, S) and to earthquakes of any depth. These generaliza-
tions have to be made on the basis of equation (1). Later, m was called the
'unified magnitude' by GUTENBERG.
However, the three scales do not agree with each other, which means
that one and the same earthquake will have different magnitude values on
the different scales. Relations between the scales have been deduced, which
make relatively simple recalculations, from one scale to another, possible
(Fig. 46):
m = 1.7 +O.8ML -O.OIMI}
(3)
m = O.56M + 2.9
The relation between m and M expresses an interesting result, which
has emanated from magnitude studies. The relation implies that for earth-
quakes at normal depth the surface waves become more and more important
in relation to the body waves the bigger the earthquake is. As seen from
the relation (3) and Figure 46 the slope is greater for the M-scale than for
the m-scale, and the two scales intersect at m = M = 6.6. This circumstance
is explained by the fact that in larger earthquakes the focal mechanism is
more extended both in space and time, and this has a favourable influence
on the generation of longer-period surface waves.
ML has only a rather limited application, but is still generally used
in studies of regional events. The use of any well-calibrated short-period
seismograph system and conversion into the amplitudes of the standard
Wood-Anderson seismograph have enlarged the applicability of the M L -
scale. Best results are obtained if such measurements are combined with
regionally determined attenuation values, preferably by spectral methods,
instead of using those derived for California. An example of this procedure
is presented by BATH et al. (1976), where an ML -scale is developed for Fen-
noscandia. The other magnitudes (m and M) have a much wider application.
One complication arises from the fact that some stations use only the
M-scale, others only the m-scale, and it is not always obvious which scale
has been applied.
But there are also a number of other complications, depending par-
ticularly on the way in which the measurements on the seismograms are
made. One important reason for discrepancies is illustrated by a schematic
case in Figure 47. While the magnitude scales were defined by GUTENBERG
as applicable to the maximum amplitudes in each wave group (correspond-
ing to Am in Fig. 47), some seismologists are instead in the habit of measur-
ing the very first swing (AI in Fig. 47). Obviously, in the two ceses shown
in Figure 47, the magnitudes would erroneously be the same, if AI is
used, whereas the magnitude would be 0.7 units higher in the latter case
if the maximum amplitudes are measured (assuming other parameters to
be the same in the two cases).
Another cause of discrepancies arises when surface waves are also
used for magnitude determination for focal depths in excess of normal.
We have seen in Chapter 3 that the surface waves are best developed for
a surface focus and that they decrease with increasing focal depth. How-
ever, consistent values of magnitudes can be obtained from surface waves
for focal depths down to about 100 km, provided the magnitudes are cor-
rected by +0.008 h, where h is the depth in km. Even for such a relatively
small depth as 60 km this magnitude correction amounts to no less than
+0.5. However, there are quite large variations in this correction from
118 Source Parameters and Their Determination
~- TIME
TIME
a
b) Evaluation of m = logy+f(.1, h) for PZ, T=1 sec and h=normal.
The main significance of the magnitude lies in the fact that it permits
a classification of earthquakes based upon the energy released. The mag-
nitude has a simple relation to the total energy of the seismic waves (E, ergs),
which has been released in an earthquake:
and E. This has involved great problems, and ever since the 1930's a great
number of different relations of this kind have been proposed, some of
them deviating strongly from each other. Equation (4) seems to represent
one of the more reliable solutions. In the derivation of an equation like
(4) M is determined according to the usual method, while the energy E has
to be computed, usually by an integration over the whole wave train under
study, i.e. both in time and space. Such computations have been made
both for surface waves and for body waves. Modern techniques (especially
magnetic-tape recording combined with the use of electronic computers)
have considerably simplified the extensive integration work. In comparison
with energy calculations from seismograms, energy determinations from
geodetic measurements in the epicentral area have only a limited applica-
tion. In fact, only relatively few earthquakes permit reliable measurements
in the field; moreover, it is unclear what the relation is between such mea-
surements and the total seismic wave energy E released.
From Tables 7 and 8 it is clear that earthquakes cover an enormous
energy range. This great range can be even better elucidated by the follow-
ing comparison. An earthquake of the relatively modest magnitude M=6.8
corresponds to an energy release of about 290 million kWh, i.e. the electric
energy consumption in Uppsala during 1966 (then about 100000 inhab-
itants). An earthquake of M =8 i releases an amount of energy equivalent to
the electric energy consumption in Uppsala for 670 years, assuming the
same annual value as for 1966. And a very small earthquake, with a mag-
nitude of 1.6, would correspond to the average amount of electric energy
Table 7. Corresponding values of seismic wave energy (E), magnitude (M and m),
maximum intensity (10) and maximum acceleration (ao)'
E M m 10 Qo1
erg cm/sec 2
1 10 has been calculated from M by means of equation (10), Chapter 4, and in single cases it may deviate
significantly from values really observed.
• In order to facilitate comparison among the shocks listed, the energy of the Agadir earthquake has bee n
chosen as unit.
used in Uppsala in just one second. However, hitherto it has not been
possible to bring the enormous energy of earthquakes under control or
to extract any useful work from it.
Tallie 8 also demonstrates the practically complete absence of cor-
relation between released energy and casualties (number of deaths). In
Chapter 5 we shall learn more about destructive earthquakes.
Thanks to the ease of its determination and its reliability, the magni-
tude has rightly come to stay as a dynamic source parameter of unparalleled
significance. However, in recent time several other dynamic parameters have
appeared, notably the seismic moment Mo' This is defined by the equation
(6)
4.4 Intensity
I 0 I
II 1 II
III 2 III
IV 2.3 IV
V 3 V-VI
VI 4 VII
VII 4.5 VIII
VIII 5 IX
IX 6 X
X 6 X
XI 7 X
XII 7 X
1 Intensity values according to MSK 1964 agree with those of the Mercalli-Cancani-Sieberg Scale (1917). the
Modified Mercalli Scale (1931) and the Soviet Scale (1952).
t The seven-degree Japanese Scale is often used also in international communications issued from Japan.
It is to be observed that this is an intensity scale and not a magnitude scale.
intensities from each other. The isoseismal maps are valuable comple-
ments to the instrumental records, and they are of special importance
in engineering seismology. However, it has to be emphasized that the
construction of isoseismal maps from macroseismic observations requires
a critical attitude towards the observations and a good knowledge of the
psychology of observations. Good knowledge about the geological con-
ditions and the types of building construction used in the area is also of
essential importance. As a rule, the intensity is greater on loose ground
than on solid bedrock, an observation which has been confirmed by
instrumental records. Whether the intensity is smaller below the surface
(e.g. in mines) than on the surface, has not been established reliably_
Some observations, which seem to indicate such a behaviour, may depend
on different ground conditions (solid rock in the mine as compared to loose
ground on the surface). Especially for use in engineering seismology, ins-
trumental records of accelerations are of indipensable value and more
reliable than direct observations alone. Such recording is also performed
continuously in some earthquake areas, where damaging effects are ex-
pected.
Some difficulty in evaluating macro seismic observations may be
Intensity 125
caused by the fact that several non-seismic phenomena give rise to effects
similar to those of earthquakes. One example is provided by the sonic
booms from airplanes. These booms are not recorded by seismographs
except in the immediate vicinity of the phenomenon. Crack formation
in loose material due to frost action is another example. This may occur
during cold winters, especially in connection with rapid temperature varia-
tions. The shaking thus produced may remind one of a small earthquake,
but since such events are limited to loose material, they are not recorded
by seismographs. Landslides are another phenomenon restricted to the
soil and other loose material and are generally not recorded by seismo-
graphs.
By means of an isoseismal map, it is possible to calculate charac-
teristic quantities for an earthquake, such as the position of the epicentre.
the depth of the source (hypocentre) below the earth's surface and the
released energy. On the basis of the macroseismic observations, the epi-
centre is the central point where the maximum intensity has been observed.
126 Source Parameters and Their Determination
It has been found quite often that epicentres determined macro seismically
and instrumentally do not agree exactly. The deviations can depend on
inhomogeneities in the geological structure. In addition, an instrumentally
determined epicentre (from initial P-wave readings) is generally located
at the point where the fracture starts. In extended fault systems this point
can deviate considerably from the area with the highest intensity. For a
source located at some depth, the decrease of intensity outwards from the
centre is evidently much slower than for a source at or near the surface.
This provides a possibility to estimate the focal depth from the rate of
decrease of the intensity from the central area outwards. After this, the
released energy can be estimated. It can be formulated as a function of
any two of the following three variables: the maximum intensity (10 ), the
average radius of the macroseismic area (r,), the depth of the hypocentre
(h). See Figure 49. It is of essential importance to notice that knowing
only one of these quantities is not sufficient to define the released energy
in an unambiguous way; instead two of these quantities are required in
the general case. It is natural that an earthquake at a certain depth and
with a certain macroseismic area has a certain energy. For instance, the
maximum intensity alone is not sufficient to define the energy, because
for a given maximum intensity the released energy has to be greater, the
deeper the source is located.
The following equation gives at least approximate information about
the relation between the three quantities mentioned:
= 3 log rl+h
2
10 -2 h2 (8)
Seismic Effects on Structures 127
important role in such research. Its practical aspects are clear and do not
need any further explanation.
The ground motion during an earthquake, especially in the epicentral
region, is a complicated phenomenon. A building or any other construc-
tion also represents a dynamically complex structure. We then understand
that great complexities are involved in the interaction between the two.
The motion of the ground will induce oscillatory stresses and strains in a
structure. The characteristics of the vibratory motion of a structure will
depend on the characteristics of the ground motion as well as on the pro-
perties of the structure, such as its size, shape, mass, rigidity, damping,
etc. As a rule, it is the weakest parts in a structure that will be damaged
first. Therefore, it is considered most suitable to avoid weak parts, i.e. to
make the structure as a whole of similar strength throughout, sufficient
to withstand expected seismic forces. In this way, concrete structures or
reinforced buildings can be made to withstand even very large seismic
forces. Flexible joints between walls, floors and roofs are of great help, and
are exemplified in wooden structures, but are also a general principle follow-
ed in the design of any earthquake-proof structure. Resonance effects have
to be avoided as much as possible. In case some choice for the location of a
building is given, then experience from California has demonstrated that a
location on as solid rock as possible should be selected (this being the pri-
mary request), then should in addition active faults be avoided. The re-
quirements on earthquake-proof constructions have recently become much
more severe than earlier. This is true especially in the design of nuclear
power plants, for which earthquake effects are considered even in countries
(like Sweden) where they were never before taken into account. Structures
of considerable horizontal extension, e.g. larger bridges, dams, pipelines,
are also sensitive to slow motions in the ground, such as in the vicinity of
active faults, which has to be considered in their location and construction.
In the following we shall give a brief outline of calculations on the
basis of which earthquake-resistant buildings may be constructed. The
mathematics used is rather simple, and the main difficulty derives from
lack of accurate knowledge of numerical values of all (;ontributing factors
to be applied in any special case. The forces that an earthquake may exert
on a structure can be written as follows:
Seismic Effects on Structures 129
where B assumes values between 0.6 and 0.8 and T is the period of the
vibration.
Table 10. Numerical example of the calculation of the seismic coefficient s for
T=0.5 sec, P=0.75 and,,= \; compare equation (13).
And the vertical force has approximately the same size as the horizontal.
Such calculations may serve as a guide on how to construct a building or
other structure to withstand the seismic forces. Obviously, there is quite
a number of considerations which have to be taken and which are adequately
summarized in the expression (13) for the seismic coefficient. Detailed
and accurate evaluations no doubt require close cooperation between spe-
cialists within various branches, such as geology, seismology, structural
engineering, etc. Developments similar to those given here apply to any type
of mechanical vibrations and their influence on structures. For example,
the modern problem of traffic vibrations can be treated in. an analogous
way, then putting R= 1 and considering that dominant periods T are gene-
rally short and that structural effects are of an integrated fatigue nature.
I. Not noticeable: The intensity of the vibration is below the limit of sen-
sibility; the tremor is detected and recorded by seismographs only.
II. Scarcely noticeable (very slight): Vibration is felt only by individual
people at rest in houses, especially on upper floors of buildings.
The MSK Intensity Scale of 1964 131
Chapter 5
struction and casualties, but is not useful from a purely scientific viewpoint.
There are numerous descriptions of earthquake effects to be found
in the literature, also there are a number of summaries of such descrip-
tions. Older literature becomes more and more unreliable the farther back
one goes. There are frequently gross overestimates of the number of deaths,
which also happens from time to time in the very first news from an earth-
quake disaster. Official statements from governments and rescue organiza-
tions are generally reliable. In addition, older literature often dwells upon
secondary effects, while primary effects - such as observations of fault dis-
placements - do not attract the same attention. A thorough knowledge of
earthquake effects and a sound criticism is necessary in using this literature.
On the other hand, it is often questionable to what extent surface observa-
tions of fault displacements are representative of focal mechanism, con-
sidering that for most shocks with observable effects the hypocentre is
20--30 km deep.
The effects of earthquakes on buildings and other constructions, such
as bridges, dams, etc., have been studied extensively for many earthquakes,
e.g. San Francisco in 1906, Tokyo and Yokohama in 1923, Kern County,
California, in 1952, Chile in 1960, Skopje in 1963, and others. Nowadays,
usually every earthquake which has caused damage is carefully investigated
on the spot, to a great extent thanks to special missions sent out on the
initiative of UNESCO. The experience gained by such studies is naturally
of great importance in formulating any building codes. From the engineer-
ing side it is frequently and correctly emphasized that a greatly expanded
network of strong-motion accelerographs is needed throughout the seismic
regions of the world.
Experience has shown that it is practicable to build in an earthquake-
proof way. Buildings and other constructions, specially built according to
building codes and located on good ground (preferably solid rock), have
been able to resist the stresses which arise even in the strongest earthquakes.
The examples are still few but nevertheless of the greatest significance,
as they point to the possibility that earthquake disasters to buildings can
be avoided. The additional costs required in earthquake-proof building are
in general considerably lower than the costs for rebuilding. However, hitherto
some form of building code has been applied only in about one-third of
the approximately 70 countries, where earthquake risk exists.
Earthquake Catastrophes 139
displacements 7 m, vertical
1 m, especially at Lituya
Bay and Yakutat Bay, mo-
tions conforming to San
Andreas' system. Landsli-
des, sand blows, fissures,
extensive minor faulting.
April 24 Iran: Lar (Girash) 5.9 450 Damage almost total along
a very narrow band right
through Lar. Larger de-
struction to buildings on
the alluvial plain (where Lar
is located) than to those on
surrounding rock.
May 22 Chile: 36° S-48° S 8.3 5 700 Active area 1600 km long
and 160 km wide. Fault mo-
vement began in the north
and progressed southwards.
Both horizontal and verti-
cal displacements. Floods,
tsunami, volcanic activity.
This is a repetition of events
in 1835-7 in the same
area.
1963 July 26 Yugoslavia: Skopje 6.0 1100 Like Agadir 1960, this
shock occurred at shallow
depth right under a city;
80 % of the buildings in
Skopje destroyed; 295 after-
shocks felt up to August 15,
1963. Skopje had earlier
been destroyed by earth-
quakes in 518 and 1555.
1964 March 28 Alaska: Anchorage,
Seward 8.5 131 Most destructive earth-
quake in this area since
1899. Fissures in Ancho-
rage, 10 m deep and 15 m
wide. Horizontal and verti-
cal displacements of 6 m.
Aftershocks within an area
of 900 km X 300 km. Fires,
tsunamis (Yakutat Bay 8 m
high, Kodiak Island 3 m,
Easter Island 30 cm), minor
volcanic activity. Seiches,
2 m high, on the coasts of
Texas and Louisiana.
10 l-
E M~7
- 10 24 ergs
-
5
.--- MEAN
~
~
i---
I--- .---
I---
1968 69 70 71 72 73 74 75 1976
o
~ N M~7
01-
.---
MEAN ~
0 .----
f-- ..--
I--- f--
I--- Fig. 50. Total number N and cor-
responding energy release E for
10 I- all earthquakes with magnitude
M~7 for the years 1968 to 1976.
The respective mean values for the
1968 69 70 71 72 73 74 75 1976 same period are indicated by hori-
o zontallines.
particular does not mean very much, even though it will be of the greatest
interest to companies issuing earthquake insurances. The maximum inten-
sity (/0) is naturally a useful quantity, but frequently not enough. A comp-
lete description has naturally to take due account of all details and as far
as possible express the effects numerically.
Concerning the strength of an earthquake, this is judged by the
general public from the damage that the earthquake may have caused.
This is quite clear, and only the seismologist, who also has the opportunity
of examining his records, may arrive at another opinion. If we look upon
earthquakes as expressions of the dynamics of the earth then the magnitude
or the released energy are the only reliable measures of the strength of an
Frequency and Energy of Earthquakes 153
Table 12. Number and energy of earthquakes for the earth as a whole.
8.5-8.9 3 156
8.0-8.4 11 113
7.5-7.9 31 80
7.0-7.4 149 58
6.5-6.9 560 41
6.0-6.4 2100 30
about two shocks every minute. Of these only the stronger ones, numbering
a few thousand per year, are recorded world-wide by seismograph networks
(see Chapter 2).
The number of earthquakes of magnitude 6.0 and over occurring
on the earth as a whole is summarized in Table 12, using 10 years as a
reference time unit. For magnitudes M?E7.0, the numbers given are those
actually observed for an interval of 47 years (1918-64), based upon a
compilation by S. J. DUDA at the Seismological Institute, Uppsala. For
magnitudes M <- 7.0, the numbers in Table 12, have been obtained by extra-
polation, and therefore they have to be taken with some reservation.
The number N of earthquakes with the magnitude M can be expressed
by the following simple formula:
where a and b are positive constants. Referring N to une year and a mag-
nitude interval of 0.1 unit (M±O.OS), we find for the above-mentioned
material for 1918-64 that a=8.73 and b=l.lS (Fig. Sl):
2.5
l'
X
log N'
1.5 X
X@
X
X
@
1.0
~,
X
0.5
@
~
X X
M
o L--_ _ _- L _ ---*---X~
7.0 7.5 8.0 8.5 9.0
Fig. 51. Earthquake frequency for the earth as a whole for the interval 1918-64.
In a certain way the two effects, expressed by (2) and (3), counteract each
other. As we proceed towards lower and lower magnitudes, the energy of
each shock decreases according to (3), but at the same time the number
of shocks increases according to (2). In order to see what the net result
will be, we combine (2) and (3) into the following formula:
log EN = 20.97 to.29M (4)
where EN is the energy release, expressed in ergs per year, corresponding
156 Earthquake Statistics and Earthquake Geography
usually frequent and there is no need to wait a long time until statistically
sufficient material has been accumulated. When this is available, it is possible
to calculate the constants a and b in equation (1) and then to extrapolate
this result to earthquakes of higher magnitudes. In so doing, it is generally
assumed that a and b are the same for large as for small earthquakes. In
some cases the method has been successful, for example in Nevada, where
it has been possible to reproduce the secular statistics within a short time.
But certain other results suggest that the coefficient b is not the same for
all magnitude ranges. Thus, it has been found in some cases that b increases
considerably with increasing magnitude. In such cases a simple extrapola-
tion would lead to an overestimate of the number of large earthquakes.
In concluding this section, I have summarized in Table 13 complete
data for all the largest earthquakes (magnitude M~ 8.5) which have occurred
in the interval 1897-1976. The data are based on the most reliable sources
available. It should also be noted that all data in Table 13 are deduced
from instrumental records.
Table 13. Earthquakes with magnitude M"" 8.5 during the interval 1897-1976.
Date Origin time Latitude Longitude Focal Region Magni.
Greenwi(.h deg deg depth tude
h m s km M
~
o
~::r
[
t:J
In·
LARGE s.CT
SHALLOW
EARTHQUA KES
s.g .
• WAGNITUD £ T~71' , I . ... - Il n
v.
~
Fig. 52. Geographi cal distributio n of the larger earthquak es
according to B. GUTENBER G and C . F. R ICHTER .
-
160 EarthquaKe Statistics and Earthquake Geography
3'"
0-
c:
!<
o·
"o
en
!!.
3o
0'
'"'<
30'
.. ~
Fig. 53. The mid-Atlantic rift zone and its branches, one towards the Indian Ocean and East Africa, the other towards the Easter Island Ridge
and North America. After B. HEEZEN and M. EWING.
162 Earthquake Statistics and Earthquake Geography
mid-Atlantic belt. Concerning released energy, belt 1 takes the first place,
as already mentioned, followed by belt 3, while belt 2 comes in the third
place. Outside the narrow belts mentioned the seismic activity is consider-
ably lower, but it is hardly completely absent anywhere. Relatively stable
areas are to be found in Russia-Siberia (to the north and northwest of
the mountain ranges in Asia), in Mrica (excepting the East Mrican rift
zone and northern Africa), and in parts of North and South America.
Shield areas are exceptionally stable. Examples of such areas are to be
found in Canada, Brazil, Fennoscandia, northern Asia, parts of Mrica,
Arabia, western Australia. All seismic belts are also obvious on the fronti-
spiece map (opposite the title page of this book), which shows more recent
results, obtained by computer calculation.
However, even within the three belts mentioned, the activity is not
evenly distributed. In some parts it may be very high while intervening
parts exhibit a lower activity. A numerical comparison of different regions
is presented in Table 14, based on all earthquakes with M~7.9 in the
interval 1904-64. The total energy E released in the stated interval of
time and magnitude amounted to 2.4 X 1026 ergs. Its percentage distribu-
tion over the most important areas is clear from Table 14. All circum-
Pacific areas 'together correspond to 77% of the total energy release. As the
different regions in Table 14 are of different size or of different length, I
Table 14. Geographical distribution of the seismic energy release for M'f!E7.9 for the
interval 1904-64.
have also expressed the energy concentration by dividing the total energy
(ergs) in each belt by the length (degrees) of each belt. The figures thus
obtained are clearly to some extent dependent upon the division into
regions, but still they furnish a reliable picture of the relative distribution
of the activity. The highest energy concentration is found in the Ryukyu
Islands-Japan area. The energy concentration is about half as big as this
in the Kurile Islands-Kamchatka and in the southwest Pacific Ocean-
Philippine Islands. Only slightly lower values are found for South America,
Alaska and Central Asia to Turkey. Smaller values are obtained for the
Aleutian Islands and Mexico-Central America, these two areas being
comparable. For all other areas the energy concentration is considerably
lower and it has not been possible to calculate it. For the Pacific area we
can distinguish two 'poles' with maximum activity, one located in the Japan
area, the other in Chile.
On the whole, earthquakes are located in areas with active tectonics,
formation of mountain ranges, folding, etc. The relation between earth-
quakes and volcanic eruptions is a question which is frequently raised.
In general, it can be said that earthquakes which have a direct connection
with volcanic eruptions (so-called volcanic earthquakes) are small, whereas
all important earthquakes are tectonic. Nevertheless, there is often a certain
parallelism in the occurrence of tectonic earthquakes and voicanoes, but
presumably with no immediate genetic connection between the two.
For the tectonic earthquakes there are essentially two types of tectonics
which are of significance: arc tectonics and block tectonics. Arc tectonics
is illustrated by Figure 54; block tectonics implies that blocks of the solid
earth move in relation to each other, often along rectilinear, vertical fault
surfaces. Arc structures dominate in most of the circum-Pacific earthquakes
as well as in the Alpide belt. Along the western margin of the Pacific Ocean,
the arc structure consists mostly of island arcs, while in the Alpide belt it
appears mostly as mountain ranges (as the Himalayas, the Alps, the Car-
pathians). Block tectonics dominates in certain parts of the circum-Pacific
area, especially in California and the central part of New Zealand as well
as in the Pamir-Baikal zone and along the mid-Atlantic belt, i.e. especially
where earthquakes of normal depth occur. In some areas the two types
overlap, as in Japan, Peru, the Philippines. Earthquakes of arc tectonics
are in the majority, but information about them derives mostly from in-
164 Earthquake Statistics and Earthquake Geography
o
o _0
o 00
o 0
o 0 ~
o
o 0
Fig. 54. Vertical section of a typical so-called Pacific arc in northern Japan. After B. GUTENBERG and C. F.
RICHTER.
c'o'
-
* VO L CANOES
IIIIIII ~ III LI NE OF VOLCANOES
• • • M • 6.0 .9
- 6
, • 0 M • 7.0 - 7. 7
T RENCH OM . 7.75 - 8.5
100
200
300
400
500
600
b) 0 - 2- X_
...- 0 .""..-
3+"
.., .- .- ..-- .-
100 h /
/
km
E /"
0'" ,/X ..... //+
200 N,/ M
/ /
/
o ,/
,/
X +/
300 / / I
/ I I
.., .-- 0 ,/X .....
.. +
400 ..-- ,/ L . ,/
M
0::::"_
.....
/
X- ..... 7.0 ~".... 8.0 9.0
500 "- "- ....
600
"' J
0 "'X/ '\
+
\
I / I
700
..,0 ..--X ..... + log E
~
2 3 4
Fig. 56. The depth dependence ofsei.mic activity.E=total seismic energy (unit = 10" ergs) for all earthquakes
with M?ii7 for the intenal 1918-64, N=corresponding total number of earthquakes, M=maximal magni-
tude: a) for depth intervals of 25 km (the points are connected by straight lines), b) for depth intervals of
100 km (the dashed curves suggest a plausible generalization).
169
Chapter 6
Earthquake Mechanism
Fig. 57. Part of the San Andreas fault in California. After H. BENIOFF, modified.
Solutions for Individual Earthquakes 171
ferent places, and the fault zone slopes in under the continent in the way
that the figure shows (compare Fig. 54). Later in this chapter, in the section
on the new global tectonics, we shall see that the sloping fault zone in Figure
59 has a modified explanation. In fact, both Figure 58 and 59 are nowadays
more of a historical significance in the development of our knowledge, but
their importance for present-day seismology is still so great that they cer-
tainly deserve to be described and known.
IZ I Z
Jv.P. P-
~
S S
P- P.
~
IZ
-+
Iz
A B c 0
Fig. 58. A simple model (in the horizontal plane) explaining the fault mechanism in tectonic earthquakes.
P+ =compression (anaseism), i.e. the first motion of the P-wave is directed away from the source (NW
and SE quadrants above), P- =dilatation (kataseism), i. e. the first motion of the P-wave is directed towards
the source (NE and SW quadrants above).
both along and perpendicular to the fault surface. In Figure 58 the radia-
tion diagrams are shown for P and S for the simple model illustrated in
that figure.
While earthquake parameters can be calculated, at least approximately,
from the records at only one station (Chapter 4), the determination of
the source mechanism evidently requires a combination of records ob-
tained at a great number of stations in different directions from the source.
Earlier studies of this kind were almost exclusively based on the direc-
tion of first motion of the P-wave (compression or dilatation). In more recent
studies, both P- and S-waves as well as reflected body waves and surface
waves have been used and in addition spectral methods are applied.
By including as many different wave types as possible in the calcula-
tion of focal mechanisms, we eliminate to a certain extent the lack of
observations in critical directions. In addition, such a combination provides
a much deeper insight into the properties of different waves, their propaga-
tion, etc., as all waves have to agree with one mechanism solution, when
correctly interpreted. Under the assumption that all waves have originated
by one and the same motion at the hypocentre, then the different phases
(their amplitudes, direction of displacement in the records, etc.) are evidently
not independent of each other. If we start from the record and, so to speak,
follow the waves backwards to the hypocentre with due regard to all changes
in amplitudes, phase angles, etc., which have occurred at the different
discontinuity surfaces or within continuously varying media, then all phases
should agree. The partitioning into two reflected waves (P and S) and two
refracted waves (P and S), which occurs when an elastic wave (P or S)
strikes a given discontinuity surface (in density or elastic properties or both),
is unambiguously determined. In principle, it is easy to calculate, even though
algebraically rather cumbersome expressions are met with in the general case.
However, as always when a problem is inverted we have to be careful
and check if the solution is unique. For a given focal mechanism it is a
straightforward task to calculate the radiation of different waves and their
distribution over the earth's surface. This is what we did in Figure 58 for
a simple case. But in practice, we have the inverted problem: to draw
conclusions about the mechanism from given distributions. In addition,
we could not exclude the possibility of curved fault surfaces or of curvi-
Solutions for Individual Earthquakes 175
--0
--{!)
--e
--®
--e
'®
'@
~'e
'~
~ Fig. 61. Focal sphere for a single force
couple F with moment (single dipole with
'\ \ '\
'"
'-,;:, moment). Radial displacements (P): plus
outwards, minus inwards. Tangential dis-
\'
placements (S): indicated by arrows. F =
focus, FP = trace of the fault plane, BP =
trace of the auxiliary plane, perpendicular
®-- FP
®.-. to the fault plane, PBP=pole of the auxili-
ary plane, N = null vector.
radial displacements on the surface of the sphere and the S-waves as tan-
gential displacements. These are in turn split into SV and SH. Figure 61
shows a typical case.
It is quite clear that we need observations with a good coverage of
the focal sphere in order to be able to make any reliable conclusions about
the mechanism. Unfortunately, in many practical applications only a small
part of this sphere is covered by observations (Fig. 62). For example, if
P-observations are available only for distances exceeding 10°, then only
about 1/3 of the surface of the focal sphere is covered by observations
(for 20° and 30° the corresponding fractions are only about 1/6 and 1/10'
respectively). This is important to keep in mind in any focal mechanism
solution. This deficiency can be remedied in two ways:
1. By including observations as near to the epicentre as possible (i.e.
for distances less than 10 0).
176 Earthquake Mechanism
0.4
0.2
0.1
0.05
Fig. 62. The fraction (y) of the surface of the
focal sphere that is covered by P-wave observa-
tions when these are available only for distances
~_~_ _~_ _~_ _~--L!.~ _ _~!A_o_ larger than .1 0 ' The figure has been derived
10 30 50 70 0 with an assumed focal depth of 100 km.
do not exhibit the simple distribution scheme of Figure 58, but instead
show four lobes, just like the P-waves but rotated 45° in relation to the
P-lobes (Fig. 63). Such a radiation pattern can be explained by assuming
two force couples at the source instead of just one couple. In such cases,
obviously not even the S-waves are able to discriminate between the fault
plane and the auxiliary plane. In the seismological literature much dis-
cussion can be found about the nature of mechanisms, especially if there has
been a single-couple source or a double-couple source. BENIOFF has shown
that both hypotheses are in agreement with REID'S elastic rebound theory.
Modern investigations are much more in favour of the double-couple
explanation (Fig. 63) than the single-couple one (Fig. 58).
In recent years, surface waves have also been included in mechanism
studies. By and large, it is true to say that the radiation of Rayleigh waves
is related to the radiation of P and that Love-wave radiation is related to
S. In most major earthquakes the active fault has a considerable extent
in the horizontal direction, sometimes up to around 1000 km. Observa-
tions of surface waves have been particularly informative in such cases.
We then have to imagine that the fault motion starts in some part of the
fracture zone, usually near one end of this zone, and then propagates along
the whole fracture. Obviously, by such a phenomenon the major earth-
quakes will be extended, both in space and in time with certain ensuing
complications in the seismic records. It has been found theoretically and
also confirmed by observations that surface waves which propagate in the
same direction as the fault motion will have shorter periods and larger
amplitudes than those surface waves which propagate in the opposite direc-
tion (compare the DOPPLER principle). In an application of this theory to
observations of Rayleigh waves from the Chile earthquake of May 22,
1960, a velocity of fault propagation of 4.5 km/sec was found and a fault
length of 750-800 km of the active segment. The seismologist A. BEN-
MENAHEM (Israel) has performed pioneering investigations of great impor-
tance in the application of surface waves to the study of focal mechanisms.
Hitherto, the studies have mostly concerned fundamental-mode surface
waves, and the relations between mechanism and higher-mode surface
waves largely remain to be investigated.
Another method to calculate the propagation of a fault depends
on phase relations between vertical and horizontal components of spheroidal
Solutions for Individual Earthquakes 179
free vibrations of the earth. The phase difference can have any value be-
tween 0° and 180° depending upon the wave-length, the fault length and
the velocity of fault propagation. For the same Chile earthquake as just
mentioned it was found in this way that the fault propagation velocity
was 3-4 km/sec and the active fault length was 960-1280 km. Taken to-
gether, this means that the total duration of the source phenomenon was
not less than about 5 minutes.
These theories are of great significance, especially as they permit a
calculation of fault length and fault propagation velocity also in cases
which are not accessible to direct observation on the spot. Direct observa-
tions have revealed displacements of the order of a few centimetres in
small shocks up to about 14.5 m (Yakutat Bay, Alaska, 1899). The horizontal
extent of the active fault zone also varies from a few metres in small shocks
to around 1000 km in the largest. The greatest extent of directly observed
displacements was encountered in the San Francisco earthquake of 1906
which could be followed over a length of 330 km. Most of the large fault
zones are located around the Pacific Ocean, but with the exception of
the San Andreas fault zone in California they are mostly located off the
coasts and thus in general evade direct observations.
In interpreting mechanism solutions, we have to remember that our
immediate results concern the orientation of the fault surface and the
direction of motion along this surface. The next step is to try, by means
of these results, to learn something about the stress system, which has
caused the rupture. These stresses must be considered to have a more
primary importance than just the fault motion.
However, the relations between fault motions and earthquakes have
in recent time been subjected to renewed scrutiny. Some seismologists
prefer to consider the primary phenomenon in the earthquake source as
a sudden volume change (for instance, caused by some phase change of
the matter), and then the formation of fracture should be of a more sec-
ondary nature. To date, this view has not gained many adherents, and it
may be true that this picture would fit better for deep earthquakes (h>
>450 km) than for the shallower ones. Particularly for the deep earth-
quakes there still remains much to be investigated in this respect.
BENIOFF reports that records on strain seismographs in South America
of two deep South-American earthquakes suggest that the elastic rebound
180 Earthquake Mechanism
40 ~--+----+----~~4-~~----r---~--~----+----+~~r-~
Fig. 64. Directions of the principal horizontal stress around the Pacific Ocean. After G. J. LENSEN (1960).
ro'N ~----------~~------------~
SO'N
40
40
20
Fig. 65. Earthquake belts in continental
20
Asia and a hypothetical explanation (the
arrows are supposed to indicate possible
slow motions in the earth). Earthquakes
with M?:o7 for the interval 1897-1964
according to S. J. DUDA'S table are located
within the shaded areas.
80
modern theories of global tectonics integrate both the rift system and the
arc structures into one world-wide structural pattern.
In this connection, it should be mentioned that the planes, along
which deeper earthquakes (h >300 km) are located, exhibit dip angles of
about 55°-60° (Fig. 55). Two American scientists found in 1968 that these
planes are tangential to the outer core boundary, provided they are suffi-
ciently extended. According to their suggestion, relative motions of core
and mantle generate tangential stresses which in some way are transmitted
towards the earth's surface and released there in the form of earthquakes.
However, it appears improbable that tangential stresses could arise at this
boundary between a liquid and a solid medium, especially as it concerns
stress accumulation over long periods of time. In contrast, normal stresses
could be envisaged at the boundary between core and mantle, especially
if one adheres to the hypothesis of an expanding core. Such pressures
could cause the mantle to rupture in its outer part along planes which
would be approximately tangential to the core boundary. This has been
demonstrated by laboratory experiments.
In this section, we have tried to take the step from mechanisms
of individual earthquakes to systems of continental size. We have seen that
some generalizations in this way are possible. The next step would be to
extend the system to comprise the whole earth, i.e. to try to explain all
the different expressions of the earth's activity, in the form of earthquakes
and otherwise, as due to one and the same underlying reason. Efforts in
this direction have not been lacking. In the most recent years a theory
of ocean-floor spreading has played a very great role in the discussions,
but still much remains to be investigated.
Fig. 66. Relative motions envisaged in the ocean-floor spreading hypothesis. Ridge structure in the centre,
arc structures to the left and right. After B. [SACKS et al. (1968).
(called transform faults, rather than simple strike-slip faults), which mark
the lines of breakage. New material is supplied from below. The ocean
floor, which drifts away from a ridge, will proceed until it hits a continental
edge. Here it bends downwards, and creates the dipping planes well known
The New Global Tectonics 187
~
(")
::r
30' Il'
;:J
V;'
S
PACIFIC
3 0'
;II
'>1:
.) T"-
60 '
~-
120' 120' o·
Fig. 67. Relative motions (compression and extension) of six lithospheric blocks (after X. Le PICHON, 1968), Crosses denote historically active volcanoes,
open circles denote earthquakes which have generated major tsunamis.
The New Global Tectonics 189
with its rotation centre. The location of the rotation centres were deter-
mined by least-square procedures, both from the orientation of transform
faults and from the known rates of displacement. The centres given in
Table 15 are based upon the orientation of the fracture zones. We note,
for instance, that the centres of rotation both for the Atlantic and for the
North Pacific are relatively near to the southern tip of Greenland. These
results can be interpreted in terms of a number of blocks which move
relatively to each other. As a first approximation, LE PICHON discriminates
between six major blocks, as indicated in Figure 67. Their relative motions
are given in Table 16, which is a summary of a more detailed tabulation by
LE PICHON. The system of motions forms a world-wide pattern in which
all motions are interrelated. The pattern obtained permits extrapolation
back in time, which reveals the nature and extent of continental drift in
Table 16. Differential movements between blocks (Fig. 67) after X. LE PICHON (1968).
Eurasia-Pacific -8.7 1 6
Australia-Pacific -5.0 5
America-Antarctica -4.3 8
America-Pacific -5.9 3
America-Eurasia -1.5 2
Africa-Antarctica +2.1 2
Australia-Antarctica +6.1 2
Africa-Eurasia -2.1 4
Australia-Eurasia -5.1 5
Fig. 68. Direction of relative horizontal motion, as derived from earthquake mechanism studies. Double
lines mark the world rift system, heavy lines mark arc structures, thin, single lines mark major transform faults.
After B. ISAcKs et a!. (1968).
occur. The fact that earthquakes are not found to occur below about
700 km is therefore not explained by any world-wide property of the earth's
interior, but is just a more localized feature, bound to island arc structures.
But how do stresses arise in the descending lithospheric slab at the island
arcs? If the surrounding medium at the continental margin is rigid enough,
the relative motions can generate stresses which can be stored for some
time. But with a non-rigid surrounding medium and with a uniformly
moving lithospheric slab, any stresses would hardly be expected, except
where the slab bends or meets resistance. Alternatively, it has been speculated
that slow phase changes in the slab may generate stresses that ultimately
produce fracture. The problem of the mechanism of deep earthquakes as
compared to shallow ones is no doubt related to the present picture of a
down-going slab. Moreover, it is to be noticed that in the motion of the
lithosphere from the ridges towards the island arcs not enough stresses
are generated to produce earthquakes. The oceanic areas outside the ridge
systems are remarkably free from earthquakes, as we have seen.
There are still several problems connected with the new theories.
For instance, we need explanations of the seismic differences between the
coastal areas of the Pacific Ocean on the one hand and the Atlantic and
Indian Oceans on the other. Also the high compressional stresses found in
absolute stress measurements in Iceland must be considered in relation to
the rift tension required in the new theories.
One major question concerns the nature of the driving force which
can produce all these motions. In earlier discussions of possible continental
drift, GUTENBERG maintained that no force is known which would produce
such world-wide motions. Nowadays, the situation seems to have changed,
as well-established motions do exist, and they cannot be dismissed for the
reason that no large enough driving force is known. Instead, we have to
find this force, whatever it is. There are two main ideas which dominate
these discussions, i.e. the hypotheses of an expanding earth and of con-
vection currents in the earth's interior. It has been considered unlikely
that convection currents could be the reason, as these would then have
to be oriented along the ridge structures. Rather it is assumed that the
breaking apart of the lithosphere slabs and their motion are due to some
underlying state of stress. It has been argued that a possible reason for such
a state of stress is an expansion of the earth. However, with the known
Time Series of Earthquakes 193
Alaska, March 28, 1964 8.5 6.8 About 17 hours after the main shock
Aleutians,February 4, 19658.1 7.2 About 3.5 hours after the main shock
Japan, May 16, 1968 8.6 8.0 About 10 hours after the main shock
Kuriles, August 11, 1969 8.1 6.9 About 7.5 hours after the main shock
Table 18. Number of aftershocks per day (24 hours) counted from the time of the
main shock, according to records firstly at Uppsala, secondly over the Swedish net-
work, for some of the most important earthquakes during the 1960's. The minimum
magnitudes are around 5-5.5 on the Richter scale.
1 The days are counted from the time of the main shock: I. = first day, i.e. 0-24 hours after the main shock~
2. = second day, i.e. 24-48 hours after the main shock; 3. =third day, i.e. 48-72 hours after the main shock.
recorded from some of the most important series during the 1960's. In
addition to the numbers actually observed I have also given reduced num-
bers in order to facilitate comparisons. The column BIA gives the ratio
of the number of shocks recorded over the whole Swedish network to the
number of shocks recorded only at Uppsala. From Table 18 we can learn
the following facts:
1. The ratio between the number of shocks recorded by the whole
network (Table 2) and only at Uppsala (BIA) exhibits a strong dependence
on the location of the shocks. Most likely, the mechanism of the earthquakes
is the reason for such behaviour, along with different sensitivity of our dif-
ferent stations.
2. The decrease in the number of shocks from the first to the second
and third days shows remarkably smaller variations. Disregarding minor
differences, this decrease is about the same for all earthquake areas investi-
gated and also it is about the same at all our stations as it is at Uppsala alone.
196 Earthquake Mechanism
b)
c) ~~,~~~F~++~I
--i-~-"'-:f--+--+-I-+-
~7~~~--~-+~~ Fig. 69. Relations between the time variation of stress and of
~~O~-L~--~·~I strain: a) The time variation of the applied stress, b) and c) result.
TIME --.- ing strains for different materials. After H. BENIOFF.
Time Series of Earthquakes 197
Moreover, it was clear that the essential difference between large and small
earthquakes is not to be found in the strain (deformation per unit volume),
but primarily in the size of the volume which is in a state of stress and in
198 Earthquake Mechanism
11I
I I
-lJf I
1
i DEFORMATION
I 16 i 3
'-9'10 em
i I
_ A - Aleutian
M ~5.9
Islands
1957 -t·_-;;
I
~
: I
i
i1 8 - , _____
,
--I
I
B
-
I
Chile
M ~5.6
1960 I Jl--h I I I
I I
II
7-~---+ : ! i
~l
I
I I
I I
-1--+1
I I i
+-----1-,
§
~ 6---
~~
~ ----
~
~
fi
~
I
1 l! I I I
75--,--,' 67 65
727.0·. . 6.2 6.0 "> i
'I . MAGNITUDES B I i
L5 ----+--1 +-~!
I
1
: I
. I I
~Fli
IJ !
- -H
fi
III
1
I I
3-,
Fig. 70. Deformation characteristics for two aftershock sequences: Aleutian Islands 1957 and Chile 1960.
After M. BATH and S. J. DUDA.
200 Earthquake Mechanism
O__----------------------~r---------~~------~~---,
.-
1 \ ~"}'
\ ...()~'
~I
3
\ SECULAR ENERGY
ACCUMULATION ~I
\ ",.-.-,)
\
5 \
6 \
>- 7 \
CD \
...
It:
8
\
z
9 \
'" 10 \
\
~ ~ 11 \ PRINCIPAL EARTHQUAKE
12
\
1
ElASTIC RECOVERY
13 \
14f---.- '------,------- ,------r------r----
1813 1830 1850 1870 1890 1910 1930.01 0.1 1.0 10 100
YEAR DAYS
Fig. 71. Energy accumulation (arbitrary scale) and release. The purpose of the figure is to demonstrate the
differences both between the main shock and the aftershocks as well as between shallow and deep earthquakes.
40 x 10 25
E
ergs
35
30
25
20
15
ALL EARTHQUAKES M ~ 7
10 1897 -1976
Year
1900 1910 1920 1930 1940 1950 1960 1970
Fig. 72. Accumulated energy release for all earthquakes with M?;;7 for the interval 1897-1976. The curve
has been calculated by means of equation (4) in Chapter 4, using S. J. DUDA'S tables for 1897-1964 and the
monthly bulletins of the Seismological Institute, Uppsala, for the following years.
Many laymen have the impression that the seismic activity on the
earth has increased in recent years. The reason that many people have
such an impression is to be found partly in an intensification of the seis-
mological reporting activity, such that newspapers contain more frequent
and more detailed information on earthquakes than earlier. Another reason
is to be found in the expansion of the population to earlier uninhabited
areas which is presently going on in many eartllquake countries (for in-
202 Earthquake Mechanism
Chapter 7
, The thickness of the earth's crust averaged over the whole earth is about 10 km.
2 1 megaton = 106 tons.
0
30
0)
25
t , _ _
min
20
15 b)
10
5
Fig. 74. Travel-time curves for P for a surface focus: a) for a
A
homogeneous sphere (v p =6.0 km /sec), b) for the real earth
o 50 100·
(according to JEFFREYS and BULLEN).
The Main Features of the Earth's Physical Properties 205
in kilometres along the earth's surface). Then the travel-time graph has
to be curved even when Vp is constant.
Figure 74 also shows the observed travel-time graph for P for a
surface focus. We find that the observed graph gives lower times than the
one corresponding to a homogeneous sphere and moreover, the observed
graph is more curved than in the homogeneous case. These circumstances
can be interpreted in only one way: that the waves have propagated with
higher velocities during some part of their path, or, in other words, that
the wave velocities in general increase with depth in the earth. However,
a consequence of this is that the wave paths are no longer rectilinear, but
curved and in fact concave towards to earth's surface. And in turn this
naturally means that the wave paths are longer than in the case of a homo-
geneous sphere. Thus, we have to determine such a velocity-depth dis-
tribution that the net result (of a longer wave path.and higher velocities).
is a decreased travel time.
It is possible to solve this problem exactly and it has also been sub-
jected to extensive mathematical treatments, including solutions of integral
equations. Symbolically, we can express the problem in the following simple
way: t(LI) -vCr), i.e. from observed travel times t we calculate the velocities
v at different distances r from the earth's centre. Conversely, a number
of theoretical models have been investigated, where one starts from an
assumed velocity distribution and deduces the corresponding travel-time
curve, i.e. a problem which can be expressed as vCr) -t(LI). This curve
can then be compared with observations and the theoretical model can
be successively modified until agreement is achieved. The transformation
v-I is a direct problem with a unique solution, whereas t-v represents the
inverse problem, usually non-unique. Here, we shall be content with re-
porting the results of these investigations.
Figure 75 shows the velocities of longitudinal and transverse waves
as a function of depth below the earth's surface. Obviously, the results
according to the two geophysicists JEFFREYS and GUTENBERG are in general
in very good agreement. Of particular interest is the pronounced decrease
of the P-wave velocity from 13.6 to 8.1 km/sec at 2900 km depth, i.e. at
the outer-core boundary. In addition, the transverse waves cease to exist
at this depth. Waves which have traversed the outer core as shear waves
have still not been found with reliability. This has led to the conclusion
206 Internal Structure of the Earth
14
vs
km/sec 13
Fig. 75. The velocities (v)
12
of longitudinal waves (P, K,
/) and transverse waves (S) in
the earth's interior according
10
toH.JEFFREYs,1939 (upper fig-
9 ure) and according to B. Gu-
TENBERG, 1958 (lower figure).
h km B
,. _ - - - - - ' - - - ' - - - - - L T=wave period, h=depth
100 1000 2000 3000 4000 5000 6000 be10w earth's surface.
that the outer core behaves as a liquid - at least in relation to seismic waves.
This conclusion is confirmed by observations of the tidal effect in the
solid earth (corresponding to the ocean tides), which is too large to be
explained by an entirely solid earth, but fits well with the assumption
of a liquid core. The irregularities of the velocity around 5000 km depth
(Fig. 75) mark the boundary of the inner core. Whether the latter is solid
or liquid, is a problem which is not yet quite settled, but some facts seem
to indica te that it is solid and thus should be able to transmit transverse
waves. For his velocity values, JEFFREYS has given error limits of at most
±0.5'X), except for the depth intervals 413-1000 km and 4982-5121 km,
where the velocities are considered somewhat less reliable.
The good agreement between JEFFREYS' and GUTENBERG'S velocity
curves is an expression of the fact that their travel-time tables agree closely.
The remaining discrepancies just illustrate the uncertainties which in spite
of this may still remain. It is particularly at two levels where interesting
deviations are found:
I. In the transition zone to the inner core. JEFFREYS has here introduced
a low-velocity layer, while GUTENBERG does not consider this to have suffi-
cient basis in the observations. The difficulties become large at these levels,
The Main Features of the Earth's Physical Properties 207
which is clear when we realize that a wave spends only a small fraction of
its total travel time in any such layer. The problem about the transition
to the inner core cannot yet be regarded as completely solved. Most likely,
the transition zone consists of several layers, to some extent reminiscent
of the layering in the earth's crust, even though the transition extends
over a few hundred kilometres in depth.
2. In the upper mantle. At a depth of 413 km, JEFFREYS has a second-
order discontinuity (i.e. a discontinuity in the velocity gradient with regard
to depth), which would explain the behaviour of the travel-time curve
around 20° distance (the so-called 20° discontinuity). GUTENBERG, on the
other hand, has completely eliminated the discontinuity at 413 km depth
and replaced it by a low-velocity layer in the upper mantle (Fig. 75). It
is still not quite settled which one of these interpretations can be regarded
as the correct one, even though the idea of a low-velocity layer has got
strong supporting evidence from a number of different observations.
The close agreement between the two velocity curves for some time
led to the idea that future research would only need to improve details
in these curves, while all the main features were well known. However,
in the last few years a number of velocity curves have been proposed which
show more significant deviations from those just discussed. Above all, this
concerns the depths around 400 and 600 km where relatively strong velocity
increases with depth have been introduced (Fig. 76). These results are based
on observations at so-called array stations, where frequently not the travel
times themselves but the derivative dt/dA has been used as a basis for the
conclusions. With a dense net of stations, this quantity can be observed in
much greater detail than has been possible hitherto. In addition, early arrivals
of PP and P' P' have been observed which may seem to require sharp dis-
continuities around 400 and 650 km depth. However, recent alternative
explanations of these early arrivals in terms of scattering (Section 3.2)
have somewhat weakened this suggestion.
It should be emphasized that in the deduction of the velocities (Fig. 75),
a spherically symmetric structure of the earth has been assumed. This assump-
tion is made also in the calculation of other properties described below.
Lateral variations of a more regional nature have been neglected, although
it is known that such variations may have a considerable depth extent,
including a major part of the lower mantle. From this viewpoint we may
208 Internal Structure of the Earth
400
DEPTH
h
800 KM
1200
1600
2000
JEFFREYS-
BULLEN
2400 MODEL 77
2800
Fig. 76. Velocities of P-waves in the mantle according to array data ('model 77') compared with the JEFFREYS-
BULLEN velocities. After M. N. TOKSOZ, M. A. CHINNERY and D. L. ANDERSON (1967), modified.
only 0.042 sec to traverse the entire earth. This means that with present-
day accuracy of time measurements at the seismograph stations (about
0.1 sec), all stations would get a simultaneous recording and it would be
impossible to locate the event. On the other hand, if the P-wave travelled
with the velocity of sound in air, then it would need 11.8 hours to penetrate
the whole earth. Even in this case it would be impossible to locate most
events, because with present frequency of earthquakes it would be impos-
sible to coordinate observations at different stations. The true time for a
PKP-wave to travel 180 0 from a surface focus is 20 min 12 sec. It is evident
that the seismic wave velocities fall within a range which is favourable
from the viewpoint of measuring techniques.
Density. The next step in our study of the earth's internal constitu-
tion is to deduce the variation of density with depth. A number of density
curves has been proposed through the course of time, the presently most
accurate one by the Australian seismologist BULLEN. In deducing the
density in the earth, we base our calculations on the following consid-
erations:
1. In a complete expression, the density of matter depends on pres-
sure, temperature and chemical composition. In the calculations, the
effects of the latter two are neglected. Therefore, the method cannot be
applied to any layer where there is reason to suspect a change in the chemi-
cal composition or a phase change. Furthermore, hydrostatic equilibrium is
assumed.
2. The calculated density curve has to yield the correct value of
the earth's total mass (Table 19) and of its moment of inertia with regard
to the axis of rotation (=0.3336MR2, with M =mass and R=radius of the
earth). These quantities are well known from astronomical observations.
In addition, the density curve has to be compatible with the velocity curves.
Let us introduce the following symbols:
p=pressure and Q=density at the distance r from the earth's centre,
m=the mass of the earth inside the radius r,
g = acceleration of gravity and ')' = gravitational constant.
Then we arrive at the following four simple relations:
1. The condition for hydrostatic equilibrium:
dp
dr = -gQ (2)
14 Bath: Introduction to Seismology
210 Internal Structure of the Earth
-k =Vp--Vs
(!
2 4
3
2
(5)
d(! ym(!
(6)
dr
r2 (V~- ~ V~)
While the density of the earth's crust (about 33 km thick in con-
tinents) is about 3 g/cm 3 , the mean density of the earth is = 5.517 g/cm 3 with
an error of about 0.08 %, as obtained by JEFFREYS from determinations
of the gravitational constant. This shows that the density in the interior
of the earth is greater than in the surface layers. In applications of the
formula (6) for the radial density gradient the mass of the surface layers
is first subtracted. This mass is somewhat different in different places,
but calculating with an average crust instead has only a very slight influence
on the density values at greater depths.
In his density calculation BULLEN applied the velocity curves which
JEFFREYS had determined (Fig. 75). Starting from a probable value of the
density=3.32 g/cm 3 at 33 km depth, the density is first calculated for the
depth range 33-413 km. In this range the velocity variations of P and
S are smooth (constant gradient) and the assumption of a constant chemical
composition is assumed to be fulfilled. In the following layer (413-984 km),
where the velocity gradients are large, it is doubtful whether the composi-
The Main Features of the Earth's Physical Properties 211
10
14,---,----,---,----,---.----.,
12
k, fl
1012 dynes/cm 2
10
6
k
km
Fig. 78. Incompressibility (k) and modulus of
rigidity (,,) at different depths (h) in the earth's
0~~10~OO~-2~OOO~~3000~--4000~~-5000~--6-00LO~ interior. According to K. E. BULLEN.
and wave velocities with depth are known, it is a simple matter of calcula-
tion, using equation (1) in Chapter 3, to get the corresponding variations
of the incompressibility k and the modulus of rigidity Jl (Fig. 78). As we
know, two elastic parameters are enough to describe the behaviour of the
material in the earth in relation to seismic waves. There are other elastic
parameters than k and Jl, but they are simply related to these. From the
density variation it is also relatively simple to calculate the depth variation
of pressure and gravity (according to equations (2) and (3); see Fig. 79).
The pressure (=the pressure of overlying layers) increases monotonically
with depth, as expected, whereas the acceleration of gravity exhibits both
maxima and minima. Thus, the acceleration has a major maximum at the
outer-core boundary, where it attains the value 1037 cmjsec2 • The reason
for this apparently irregular depth variation of gravity is to be found in
the fact that g depends partly on the mass (m) inside the respective levels
in the earth, partly on the distance (r) to the centre of the earth from the
different levels. Proceeding towards the centre, both m and r decrease.
As the acceleration is proportional to mjr2, we could expect the curve
shown in Figure 79.
Recent modifications of both the velocity and the density curves,
214 Internal Structure of the Earth
P
10 12 dynes/ cm 2
3
9
cm/sec 2
9 1000
2 800
600
400
p
200
Fig. 79. Pressure (p) and acceleration
of gravity (g) at different depths (h)
in the earth's interior. According to
o I 0
K. E. BULLEN.
1000 2000 3000 4000 5000 6000
mentioned above, will entail some modifications of all curves which have
been deduced from them.
As we have seen, the physical properties we have studied so far are
intimately connected with each other and form a consistent system. The
trend of the calculations is defined by the following points:
1. Travel times of P- and S-waves: deduced from seismograms.
2. Velocities of P- and S-waves: calculated from 1 by extensive oper-
ations.
3. Density: calculated by means of2 and other assumptions and con-
ditions.
4. Elastic parameters: calculated from 2 and 3.
5. Acceleration of gravity and pressure: calculated from 3.
Revised models. In recent years, improved information on a number
of points has called for a revision of the model presented so far for the
earth's interior. The newer development has been summarized by BULLEN
and HADDON (1970), who have included the following new results in their
revised models:
1. A new value of the earth's moment of inertia (0.3309MR2 instead
of 0.3336MR2), based on observations of artificial satellite orbits.
The Main Features of the Earth's Physical Properties 215
(7)
where the factor '1 is different from unity in the earth, except for the depth
range 3600-4500 km.
5. Studies of compressibility of materials at high pressure.
6. Shock-wave experiments.
7. Observations of the earth's free oscillations.
The revised data have led to several new earth models of which one
of the most recent ones is demonstrated in Figure 80. It is especially to
be noticed that the S-wave velocity exhibits a minimum in the upper mantle
in this model. According to the newest model of BULLEN and HADDON
the density increases from 5.62 g/cm3 to 9.89 g/cm 3 at the outer-core bound-
ary at 2878 km depth and the gravity attains the value 1080 cm/sec2 at this
level.
A recent modification concerns the radius of the earth's core. A num-
ber of different observations indicates that this radius is somewhat larger
than believed hitherto. The proposed increases vary depending upon the
kind of observations, for instance, an increase of 10-30 km according to
PcP-observations, an increase of about 15 km from observations of the
earth's free vibrations and an increase of 64 km derived from some ob-
servations of P-waves diffracted around the core boundary.
Absorption. The physical properties discussed so far are based on
purely elastic conditions in the earth's interior. However, it is a well-
known fact that seismic waves are absorbed during their propagation.
This indicates that also non-elastic effects are operative. Investigations of
absorption must be based on amplitudes of seismic waves, as distinct
from travel times used above. The studies have not yet been able to
yield a completely unique picture of the absorption, but at least some
216 Internal Structure of the Earth
~ r-
141
a
12 >-
P
i
10 I 9
9
6
p
Fig. 80. Earth model accord-
2
ing to R. A. w. HADDON and
K. E. BULLEN (1969). p=pres-
sure in units of 1012 dynes/em:!,
IJ = density in g/cm 3 , 1%( = l'p)
and P(='·s)=P- and S-ve-
o 1000 2000 3000 4000 5000 6000 loci ties in km/sec. g = accelera-
Oopth (km.) tion of grrvity in 102 cm/sec:!.
The lower the absorption Y. is, the higher is the quality factor, hence its name.
The Main Features of the Earth's Physical Properties 217
2000
log Q Q
3.0 1000
/ CRUST
/ 500
LOW - VELOCI TY
2.5
+
(CArY_E_R_ _ _..J
200
h km 50
Fig. 81. The quality factor (Q) for P-waves in dependence on the depth (h) in the mantle.
Mantle Core
5000
.
4000
a
~
i.E -+_______-,
~OVa,--(:,::19::58:)_ _
3000
_-----------~ut.nb.rg
---1 1_------------
(1951)
h km
OL-----~10~OO~--~2~000~----~3*000~----~4~OO~O------J
Fig. 82. The temperature as a function of depth in the earth after B. GUTENBERG (1951) and E. A. LUBIMOVA
(1958).
The earth's crust constitutes only a very small fraction of the whole earth
(Table 19), but nevertheless it has been investigated in more detail than
any other part. The reasons are obvious. The earth's crust is closest to
us in space and hitherto the only part of the earth's interior which has
permitted direct investigation, at least partly. The possibilities for profit-
The Earth's Crust 221
able extraction of various deposits (salt, oil, minerals, etc,) have contrib-
uted efficiently to the exploration of the earth's crust. A number of dif-
ferent methods have been used in such investigations.
Deep drilling. This represents no doubt the most reliable method to
determine the structure, not only the depth to discontinuity surfaces but
also the chemical composition. However, such direct methods have a limited
application, not only due to their high cost and the strictly local informa-
tion they give but also due to their limitation in depth. The deepest bore-
holes of a few kilometres (about 0.1-0.2% of the earth's radius) penetrate
only a fraction of the continental crust which has an average thickness
of 30-35 km. Nevertheless, several relevant projects have been started, both
in the USA and the USSR. In 1957, the Americans proposed the so-called
Mohole Project, which aimed at a drilling down to the Moho (abbrevia-
tion of MohoroviCic discontinuity, i.e. the base of the crust). Since then, much
effort has been put into realizing this project, especially in oceanic areas
where the earth's crust is thinner (only about 5 km thick in typically oceanic
areas). Even if it has not yet been possible to reach the Moho the dril-
lings have been of very great value, both scientifically and technically.
The Russians have started similar projects at a number of places in the
USSR, all of them on continental structure and aiming at bore-hole depths
of 15 km.
Even though bore holes do not reach very far when we are concerned
with the earth's internal structure, they would still be sufficient to penetrate
into the earthquake foci at several places. Many shocks occur at depths
less than 5 km, and there a drilling could reach the focal region. In combina-
tion with stress measurements in situ such an undertaking would no doubt
give very valuable information.
Apart from drilling experiments, all methods for investigation of
the earth's crust - as well as the rest of the earth's interior - are indirect, i.e.
they are observations on the earth's surface of various phenomena (seismic,
magnetic, electric, gravimetric, etc.) which depend on the conditions in the
interior. Obviously, indirect methods often involve great difficulties of inter-
pretation, and discussions of such problems occupy the geophysical litera-
ture to a very large extent.
Near earthquakes (Ll <: 10°). Records of near earthquakes have been
used extensively for the study of crustal structure, ever since the birth of
222 Internal Structure of the Earth
Vp km/sec '2=147
6 7 8 tl=71~O 1 200 300
(1)
Fig. 83. Wave propagation paths and travel-time curves (quantitatively calculated) for an assumed velocity
profile. CC = Conrad discontinuity, MM=Mohorovi~ic discontinuity.
FX )
Pg: t=--
p*: t=
VI
FA+BX
+-
AB I'-
( (10)
VI V2
FA'+B'X A'C+DB' CD
Pn: t=
VI
+ V2
+--
V3 J
I,.
sin ii
VI v2
sin i~ sin i2
I
--=-- (11)
VI V2
sin i2 1
V2 V3 J
\ (1). (2) = Pg
8 \ (3) = p. (1)
\(4) (4) = reflection at MM
~
\ (5) = Pn
6 \
\
t-..L\
6.8 "
'-
4
sec (4)
(3)
2
Vp km/sec (1)
7 8 F 300
Fig. 84. Wave propagation paths and travel-time curves (quantitatively calculated) for an assumed velocity
profile. Same notation used as in Figure 83.
20 8.78C31O
vr . .,t.
.7 •
30 30
M
8.1 •
40 C 40
M M 6.4 •
M 8.0 - 8.1
so. 50
80 60
70 z M 70
:r
« ...J 8.1 -
(/)
Z
I ~
(/)
~z I
::> UJ
h
UJ
>=
Z
« ""ct
0« DO
...J
:;: h
<Ii ....
DO
:r.., UJ
0.. U
• 6.0 - 0
km
10
water depth, i.e. where the shelf zone ends and the slope towards the deep
sea begins. Variations in the structure occur also within typical continents
or oceans. Particularly noticeable in this respect is the thickening to about
50--60 km of the earth's crust (so-called mountain roots) which is generally
observed under mountain ranges.
Surface waves. As we have seen already in Chapter 3, the velocity
dispersion of seismic surface waves depends on the structure of the earth's
crust and upper mantle. This method has been used extensively for structural
studies. Also in this case we have to deal with a difficult interpretational
problem: to translate an observed dispersion curve into some structure.
In general, the solution is not unique. On the theoretical side, a great
number of dispersion curves has been deduced for assumed structures,
which can serve as a guide in interpretations of observed dispersion (one
example is shown in Fig. 86).
LAYER Yp H
1.732Y51 Y51 ~1 H1
2 3.873v51 2.236v51 ~1
2.0
1.8 C/Ys1
U/Y S 1
1.6
1.4
The two curves in Figure 86 are limited to the left by the abscissa
cTj HI =0, and to the right by cTj HI = 00 (outside the figure). These limiting
cases can arise under the following circumstances:
1. cTjHI =0 for HI = 00, if T is finite, or for T=O, if HI is finite.
2. cTj HI = 00 for HI = 0, if T is finite, or for T= 00, if HI is finite.
kmlsfH:
'20
4.00
3.80
3.60
S " a
1 I,0 I~S~1 J~
15 / 7 J gllO /
Lan d
I 1'1 1 1/2 \
6 8 9
l1li 90 110
comer) are numbered 0- 12 in Ihe ordcro(incre asinG
vdoc it)' disp~n.ion of Raylc:1ih waVd, essenlially in Ihe period
mn,ie of 20----40 sec·. The dispersion curves (given in the upper Ti8hl~hand
FIt{. 87. A world map. show ing the distribution of the troup SANTO.
map. A . a and C denote special $t.ruClures (ocean ridges). AfletT.
Continen tality, and the correspondi ng numbers are given on ahe
The Earth's Crust 229
-------.---,,--,-,,-r,-,,',"------"---.---r-.,'~
-~
u
4.0
MANTLE
3.0 r----j<----
2.0
--OCEANIC
DISPERSION OF
RAYLEIGH WAVES
10
T sec
Fig. 88. Observed group velocity dispersion of Rayleigh waves for periods up to SOO sec. U=group velocity,
T=wave period.
prevails up to periods around 75 sec. For longer periods the wave mo-
tion will extend to greater depths, and then the relatively superficial dif-
ference between oceanic and continental structure is of little or no im-
portance.
2. The phase-velocity method. By observing the passage of a certain
phase (e.g. a maximum) across a triangle of stations, it is possible to calculate
both the velocity across the triangle and the direction of propagation for
this maximum. The velocity thus obtained is the phase or wave velocity
(Chapter 3). If the procedure is repeated for a number of maxima along a
train of surface waves, we have the material for constructing a dispersion
curve for the phase velocity. Since the 1950's, several such investigations
have been made, as in North America, Japan, Scandinavia and elsewhere.
In general, it has been found possible to interpret an observed dispersion
curve in terms of a certain average thickness of the crust within the area
occupied by the station triangle: The Scandinavian measurements thus gave
a crustal thickness of about 35-40 km with only smaller variations between
different parts of Scandinavia.
The Earth's Crust 231
has been erased in this hypothesis. This is probably the strongest objection
that can be made from the seismological point of view. In a paper dealing
with seismic waves which have traversed the core GUTENBERG (1951) says
among other things that 'no hypothesis can be considered a good approxi-
mation which disagrees with the result that the boundary of the outer core is
very sharp'. 'Waves having lengths of 10 km or even less are reflected at
both sides of the boundary.'
3. When KUHN and RITTMANN maintain that no separation of different
elements was possible (as claimed by the iron-core hypothesis), then they
imply that an equally high viscosity, as at present, existed at the beginning
of the development when the earth was supposed to consist of gaseous
solar matter. The separation according to the iron-core hypothesis has
been compared to the process in a smelting-furnace, where iron and slag are
separated.
4. EUCKEN (1944) objected that the high densities found for the core
(10-13 g/cm 3 ; see Fig. 77) cannot be explained if it were to consist of
solar matter (mainly hydrogen), whereas they can be easily explained by
an iron core. In fact, if the earth's core should consist mainly of hydrogen,
its density would hardly be more than about 1 g/cm 3 . For comparison,
the major planets (Jupiter, Saturn, Uranus, Neptune) have densities around
1 g/cm 3 as mentioned above. It does not appear impossible that this hypoth-
esis is better applicable to them than to the earth.
RAMSEY'S hypothesis. We now pass over to the hypothesis for the
earth's core that was proposed by the British scientist RAMSEY in 1948
and 1949. Like KUHN and RITTMANN, he started from a criticism of the
iron-core hypothesis. The main points of his objections are summarized here.
1. The existence of the earth's magnetism has been taken as evidence
for the iron-core hypothesis. The elements with the strongest magnetic
properties, i.e. iron, nickel, cobalt, so-called ferromagnetic elements, lose
their magnetism when their temperature exceeds a certain value (the Curie
point): for iron 768°C, for nickel 360°C, for cobalt 1137 0c. In all prob-
ability, the core temperature is higher than the highest of these values (cf.
Fig. 82), and, moreover, the pressure influence on the Curie points seems
to be insignificant. As a consequence, it appears as if these elements can-
not be magnetic at the conditions prevailing in the core. In other words,
the existence of the earth's magnetism is no evidence in favour of the
236 Internal Structure of the Earth
difference between the two hypotheses is that RAMSEY considers the sur-
face of the inner core also to correspond to a phase change, whereas ac-
cording to BULLEN this is a chemical discontinuity. One consequence of
BULLEN'S compressibility-pressure hypothesis is that the inner core is solid
in all probability.
Some recent seismological results. As already emphasized, a clarifica-
tion of the nature of the core-mantle boundary requires cooperation from
many branches of science. Seismology established the existence of this
sharp boundary, including information on changes of wave velocities,
density and other properties at this discontinuity. But beyond this, seis-
mology is not able to yield information on the exact nature of this transition.
However, recently seismology has again been put in a position to
improve our knowledge about this boundary. And this is mainly due to
detailed studies of waves, especially of the core-reflected P-wave (PcP),
which have been made possible by use of well-controlled sources (nuclear
explosions). In Figure 75 we find that even earlier a flattening of the mantle
wave velocities had been found just outside the outer core, in some cases
even a velocity decrease. This picture can now be established with more
reliability and more detail. A. K. IBRAHIM, while working at the Seismologi-
cal Institute at Uppsala, found on the basis of PcP-observations from
nuclear explosion sources the following structural features of the core-
mantle boundary:
Chapter 8
, Telex messages were started in January, 1977; before that we sent telegrams to NElS and air-mail letters
to CSEM.
2 Up to 1959, inclusive, also an annual seismological bulletin was issued, which started back in 1904. Be-
cause of the very much increased demand for rapid reports, annual seismological bulletins have no longer
the same importance as earlier. Their function is now fulfilled by the monthly bulletins.
1969 1969
Jan. 2 (cant. ) Jan. 3 (cf'lnt. )
Sk eP 15 23 40 Urn iPP 03 24 55.6
Ud iP 15 23 08.8 iSS 03 31 45
De iP 15 22 38.9 Ud iP 03 23 41.9 C
Rhodes Island (h " N). i 03 23 44.6
iPn 03 24 56.7
" 2 Um iPKP 16 05 44.0 C
Santa Cruz Islands
De iP
iPn
03 23 33.6
03 24 45.7
(h 640 km). Iran-USSR (h " 10 km).
m = 5.9, M " 5.2 (Up,Ki).
" 2 Up
Ki
iPKP
iPKP
18 09 28.4
18 09 43.9 C
Double P with P2 - P1
" 2.9 sec in average.
iSKP 18 12 53.8 Clear Pn phases.
micr Bec
SKP Z' 0;2 1.5 II
3 Ki iP 06 48 40.7 C
Um iPKP 18 09 36.6
iSKP 18 12 40.9 II
3 Ud iP 075223.7
i 18 13 01.7 Aleutian Islands (h " N).
Ud iPKP 18 09 26.7
Snuth Sandwi~h Islands II
3 Up iP 10 10 19.4
(h 80 km).
3 Ud iP 11 31 40,0
" 2 Up
Ki
iP
iP
18 18
18 17
05.4
22.2
ipP 11 32 02.8
Um iP 18 17 41.3 C
Guatemala. h = 90 km (Ud) •
Ud if 18 18 12.1 C
De eP 18 18 28
ff
3 Up iP 11 43 03.0
Japan (h = 70 km) • Um iP 11 42 40.1
i 11 42 55.5
" 3 Ud iPKP 00 37 47.6
Tonga-Kermadee Islands
Ud e(P) 11 42 54
(h 70 km). 3 Um iP 13 07 05.9
ff
3 Up iP 03 23 25.5 II
3 Up iP 13 39 13.2 C
i 03 23 28.1 i 13 39 34.3
iPn 03 24 31.3 mier sec
mier sec P Z' 0.9 0.9
P Z' 0.2 1.3 Ki iP 13 38 20.1 C
Mx E 1.0 11 iX 13 38 24.6
Mx N 2.2 14 mier sec
Mx Z 1.7 11 P Z' 0.1 0.8
Ki iP 03 23 51.3 Sk iP 13 38 52.0 C
i 03 23 53.9 Um iP 13 38 46.2 C
iPn 03 25 04.5 iX 13 38 50.4
micr sec Ud iP 13 39 13.7 c
P z' 0.3 1.0 iX 13 39 18.4
Mx E 4.0 12 i 13 39 29.5
Mx N 3.3 17 De iP 13 39 35.8 c
Mx Z 4.6 12 iX 13 39 39.9
Sk iP 03 23 57.7 Aleutian Islands
i 03 24 00.7 (h " 30 km).
iPn 03 25 20.9 X - P = 4.4 sec in average.
Um iP 03 23 31.5 If X were pI', h would be
i 03 23 34.9 about 15 km only.
(eont.) Remarkable PZ'-amplitude
(cant.)
Fig. 89. Sample page of the monthly bulletin from the Seismological Institute at Uppsala.
248 Methods to Improve Seismological Observations
of about 0.5-1.5 sec. The P-waves no doubt constitute the most important
information from any seismic station. In addition, in this period range
conditions vary considerably from place to place, much more so than
for the longer-period transverse (S) waves or for the surface waves.
For future developments of station networks and especially for in-
stallations of more advanced seismic stations, it is of considerable importance
to accumulate knowledge offactors significant for favourable seismic record-
ing. This research, as carried out in Sweden in 1964-6, is both time-consum-
ing and expensive, and therefore if some results of general applicability
can be found, these could prove useful in seismograph station planning in
other parts of the world, without the need to repeat in every area the same
or similar investigation as in Sweden.
The records of the temporary stations are measured by visual in-
spection in the same way as the records from the permanent net. Accord-
ing to my experience, the number of events (usually the number of P-
phases) recorded in any given time interval is a good measure of the signal
sensitivity. Exceptions to this rule will occur when some special source
dominates at one of the stations, but is not observed at other stations.
Avoiding such instances, the numbers in Figure 90 represent the sensiti-
vities at the various stations. They mean that while, for example, Uppsala
has recorded lOO events, Kiruna recorded 86 on similar type instruments,
etc. The figure well serves the purpose of illustrating the method and it
gives at any rate preliminary results. Figure 90 also includes readings from
stations in Finland and Norway, as obtained from their seismic bulletins.
In Figure 90 I have also tried to generalize the results, as shown
by the sketched curves. The sensitivity numbers in Figure 90 can be con-
sidered to be influenced by a regional factor and a purely local factor.
The generalization made has sense only if the regional factor dominates
over the purely local influences. The latter could be traced only with a much
denser network of stations.
We have to think of the curves in Figure 90 as being a result of
several effects of which the most important are the following:
1. Geological structure: the eastern side of the country (mostly on
granite) exhibits higher sensitivity than the western part (mostly on gneiss).
2. Distance from oceans: the greater the distance, the higher is the
sensitivity.
Signal-Sensitive Localities 251
'O~~~~~~~~~'~
1 -r-r_,~a~~~~~~~~~3T&'~
E ~
"N 'N
The source of seismic noise is to be found on the earth's surface, and this
is true for the whole period range of microseisms, from those of periods
above 20 sec down to local noise with periods around 0.1 sec or less. In
addition to looking for quiet places on the surface, as already described,
there is another way to avoid the noise, especially that of short period,
namely by recording underground (in bore holes or in mines) instead of at
the earth's surface.
In the summer of 1965 we recorded the seismic noise in the period
range of 0.03-0.13 sec (or frequency range of 33-8 cycles/sec) in three mines
(Haggruvan, Idkerberget, Stripa) in central Sweden at different levels down
to about 750 m. The measurements were taken in competent hard rock
(leptite), i.e. not in the ore veins but in the surrounding rock. There have
been some related measurements made, especially in the USA, but these
usually concern lower frequencies and sedimentary rock. These two facts
have to be taken into account in any comparison of the results.
In every individual case we found a clear amplitude decrease with
depth. As an average of 28 individual amplitude-depth curves, we found
that the noise amplitude decreases to 25% of its surface value at a depth
of 50 m, to 13 % at 100 m, to 6% at 200 m depth, and that it is less than
I % of the surface value at depths exceeding 500 m. However, individual
deviations from these averages may be large, depending on various effects.
The amplitude-depth diagrams obtained (Fig. 91) suggest an interpretation
in terms of body waves rather than surface waves, and the depth effect
is mainly to be explained as due to increasing distance from the noise
source.
Curve C in Figure 91 represents body waves and its equation is
e-O.262 b
a = 263 b (I)
log 0--
--+ 1Jx ~l
100
200
300~ / 1
r
~
I
400
.;
500~ A x
Fig. 91. Observed amplitude-depth variation (crosses)
60+ and theoretical curves: A for Rayleigh waves of O.100sec
period in a homogeneous medium, B the same as A but
700~ for 0.15-sec period, C for body waves. The unit for the
L--.L._--L-_-'--_---L_ _'--' amplitude a is 0.1 mJl.
The discussion so far has concerned the selection of the best sites on the
surface or underground. Another important method to improve sensitivity
is to combine records from a number of seismometers spread out over
an area, instead of using just a single channel. The idea is not new. Record-
ings by means of a profile of seismometers (geophones), usually ranging
from a few hundred metres to a few kilometres in length, have been used
for many years in field investigations of crustal structure and seismic
prospecting. Sometimes, an areal distribution of the seismometers has also
been applied, in addition to straight-line profiles. The new development-cal-
led array stations, i.e. an array of seismometers-dates from about 1958
and has been inspired by the interests in nuclear explosion detection by
seismic methods. The profile or array technique is not new, but its applica-
tion to continuous recordings of distant events is new.
The seismometers in an array form a regular geometric pattern, which
facilitates the combination of the different channels. At present such arrays
exist in the USA, Canada, Scotland, India, Australia, Norway, Sweden
and in a few other countries. Figure 92 shows the geometrical pattern in
some of the American arrays. The largest arrays so far built are LASA
(=Large Aperture Seismic Array, Montana, USA) and NORSAR (=Nor-
wegian Seismic Array, Norway), both consisting of over 500 seismometers
at their maximum activity.
The advantage of an array compared to a single channel is that it
makes use of the combination between the different channels. For example,
by adding up the individual outputs in an array with variable time lags
between the different seismometers, we get constructive interference for
the P-wave at a certain lag and constructive interference for the S-wave
Array Stations and Special Methods of Analysis 255
BMSO
Up
Ki
Sk
Gb
Ko
similarity from one station to another, even if the mutual station distances
are much shorter than in Sweden. This is explained by the complicated
structure of California with a much dissected crust, also by sediment layers
of varying thickness and structure. The Swedish stations, on the other
hand, are located on more homogeneous ground, and application of array-
processing techniques to this network has been successful.
Figure 93 shows a good case with excellent correlation between the
short-period P-wave signals all over the Swedish network. Records at the
Swedish stations generally exhibit the signal resemblance required by the
array methods. By combination of the records (usually a summation after
appropriate time shifts) the signal-to-noise ratio can be increased by a
factor of about two and the accuracy of the P-readings is improved. Also
the direction of the first motion of P (compression or dilatation; see Chapter
17 Bath: Introduction to Seismology
258 Methods to Improve Seismological Observations
property. Then, this property can be made the basis of the filtering tech-
nique.
This is the reason why we distinguish between different filtering tech-
niques, such as velocity filtering and frequency filtering. Velocity 1iltering
depends on different velocities of the wave trains which should be separated,
for instance a seismic body wave (P) and microseisms. Frequency filtering
may be more common and depends on different frequencies of the wave
trains to be separated.
Figure 94 shows a striking example of frequency filtering. Higher-
mode surface waves are usually seen as short-period waves riding on the
longer-period surface waves of fundamental mode (upper part of Fig. 94).
By applying a band-pass filter which permits waves of periods 6-15 sec
to pass but which excludes other periods, it is possible to isolate the higher
modes from the given record (lower part of Fig. 94). Obviously, the ac-
curacy in measuring higher modes from the filtered record is far better
10.-~-.----.---------~
5 NEVADA
JAN 19
101-----+---\-+
101-----+--\\·
w5
Cl
:::l
~ 10 t--~'_'<''''
a.
~
«5
w
>
~ 10 f--::,,",,'---+--- \\
--'
w
0::5
101--11-+\
101-----+--- \--iI
101-----+---\·+-1 \ - + - - \
5
Fig. 95. Amplitude spectra of short-period vertical-compo-
nent P-waves recorded at Swedish stations (Table 2) Cram an
1 15 2 underground nuclear explosion in Nevada on January 19,
FREQUENCY CYCLES/SEC 1968 (Table 30). After O. KULHANEK (1971).
that more progress in this field can be made if the correction is separated
into a station correction and an epicentre correction (just as for magnitu-
des; cf. Section 4.3). So far the same travel-time tables are applied for the
whole earth (except for regional studies). Information, especially from
records of some nuclear explosions, has shown that regional corrections
are needed to the presently used travel-time tables, of the order of - 2 sec
but different in different regions. Therefore, the results suggest that regional
262 Methods to Improve Seismological Observations
travel-time tables, different for different parts of the earth, will be the next
development in this field.
Other geophysical observations, for example of gravity, are always
reduced in some way before being studied. This is not the case with arrival
times of seismic waves which are used directly, apart from exceptional
cases when station corrections are applied. One could think that it would
be better to use arrival times reduced to the Mohorovicic discontinuity at
the base of the crust. The great difficulty is that before such a reduction
can be made with confidence we need to have a good knowledge of the
crustal structure under each station.
Table 22. Presently operating seismograph stations according to the list published in
March 1974 by the U.S. Geological Survey.
180·
EQUATOR
if~____~3rOo____~6if~____,90~O_____U~OO_____1,~~O____~180'
DISTANCE (D)
!;OO
N=320
w
u
100 zoS: N=81
~
(/)
(5
Z
%
!::
~
(/)
z0
;=
oS:
.....
II)
I.&-
0
Fig. 97. Some numerical in-
10 formation on world-wide sta-
a::
w
CD cos a. ::
1 -2tO"2 t tion networks with equal spac-
ing. The assumed grid is shown
::E 2(I.tO"21-> in inset in the upper left-hand
:::>
z 2'" (1 - cos 0 )
corner. where there is one sta-
n = 3(11- '11"
tion (marked with a dot)
within each equilateral tri..
angle. x = station spacing, n =
N :: the number of stations within
distance D from any point on
the earth, N = the total number
!!2- _N_ of stations and lIolA =the
A - 510
number of stations per million
km 2 •
MAGNITUDE (m)
4.0 50 6.0
z
0
~~r
100
g>-
;;::
t
Fjg. 98. Empirical relation be-
tween magnitude of explosions and
SHADOW ZO NE
0.01 RANGE maximum recording range for
short-period P-waves (excluding
a 20 40 60 80 100 150· array stations).
part of the energy, at least for shallow-focus earthquakes, and thus travel
farther. Ten to twenty stations, well selected considering their spacing
and their quality, would be a very satisfactory net for magnitude deter-
minations.
Even though a scheme as outlined here with its numerical evaluation
has to be kept in mind in any planning for network developments, it does
not represent more than a first approximation. Going a step further, it is
very important in choosing locations of new stations to take due regard
to the noise background. New sites should preferably be tested before
expensive installations are made. This means in practice that it will not
be possible to use a scheme like the one outlined here for more than a first
orientation of station locations. Tests on each spot will indicate their suit-
ability, and as a consequence we have to consider shifts of the positions
up to the same order as the station spacings. In addition, only experience
based on recordings will define the lower magnitude limit which can be
reached. Also, in practical applications, economic considerations often dic-
tate possible network developments.
Although it will certainly take many years before we have a homogene-
ous world-wide station network, considerable progress is already under
way in this direction. On the technical side we note very promising tests
as mentioned above. On the organizational side, we note that among
others UNESCO has taken an interesting action in such matters. Since
a very important part of the project concerns international territory (the
large oceans), the active cooperation of an international body like UNESCO
would be very significant.
ficance for many studies and applications, for example, the modern hypo-
theses on global tectonics as well as in engineering.
The increased precision is based upon a favourable cooperation be-
tween a number of factors, of which the following are most important:
I. A better distribution of stations over the earth.
2. Installation of special array stations.
3. Increased time accuracy, by general use of well-controlled quartz
crystal clocks.
4. More accurate travel times, especially of P-waves, derived from
records of nuclear explosions.
5. Application of large computers in the evaluation of the obser-
vations.
In the present section, I shall limit the discussion to points 1 and 2
above. Even though advances have been made in these two respects, much
still remains to be achieved. In the preceding section, I advocated the idea
of homogeneous networks over the globe. Here, [ will apply the ideas
to the development of a world network of array stations. I will base my
proposal upon already existing array stations rather than suggesting a ho-
mogeneous network (which may be unrealistic).
A world network, preferably as homogeneous as possible, is required
for numerous seismological studies. I shall here limit the discussion to
the determination of source parameters, especially hypocentre coordinates
and origin time. With regard to this problem, we can summarize com-
monly used methods of determination as follows, considering only tele-
seismic events:
I. Using only one station, by combination of times and of three-com-
ponent amplitudes of different waves.
2. Using a limited network, say of 5-10 stations over an area of
roughly 10 0 X 10° in extent, primarily by combination of P-wave readings.
3. Using one array station, say of 1°_2 ° in extent and with numer-
ous sensors, essentially by combination of P-wave readings.
4. Using a world network of ordinary stations, by combination of
P-wave arrival times. This refers to the determinations made at some world
centres, as for example NElS, ISC, CSEM.
Let us consider the accuracies achieved by the different approaches
mentioned. Method I can only provide a preliminary location, sometimes
272 Methods to Improve Seismological Observations
with wide margins, and its purpose is only to deliver rapid messages to
certain institutions and to news agencies. Method 2 is usually not applied
to teleseismic events, but only to regional ones. Method 3 has been practised
by several array stations, and it has become customary that each array
makes its own determinations. They have often large error margins. Me-
thod 4 is still superior to any other method and yields data of great value
to seismology.
The question now appears if there is any possibility to improve
source data information and then, in what ways. Method I has always a
justification for rapid information on epicentral area (and magnitude, etc.)
but has to be left out from the discussion of improvements. Comparisons
of methods 2 and 3 would be very informative. Some comparison of this
kind has been made; see Section 8.4. For increased accuracy these methods
alone are insufficient. In order to achieve higher precision, method 3 could
be amplified by the use of a world-wide network of array stations instead
of just one station at a time. Method 4 could be improved by extended and
preferably homogeneous networks as discussed above.
This system would include about 10 array stations with quite a good
coverage of the world. In applying this system to hypocentral location,
all information available from array stations should be used, i.e. they
should be treated as arrays and not as single stations. In other words, all
data which single array stations put into their source location programmes
should be combined for this world system. Then the accuracy of the results
should be compared with results already obtained from method 4. I have
not seen any such detailed and systematic study, and therefore I would
like to make a strong suggestion for this project to anybody with access
to primary data from all existing array stations (including calibration data)
and to a large computer.
Comparative studies of the achievements of the world net of array
stations and the usual method 4 are very significant for further develop-
ment of the whole project, for instance, if the additional array stations
(in southern Africa and Antarctica) would fill a need or not. A combina-
tion could also be envisaged of the world net of arrays and method 4,
i.e. the world net of ordinary stations, and the accuracy achieved by this
combination should be compared with either of the methods applied sep-
arately.
We have discussed accuracy of source determination by different
methods. Closely related to this problem is the sensitivity of any seismo-
graph system. The networks suggested should be compared also from this
point of view and particularly the lower magnitude limit should be speci-
fied, above which a homogeneous world-wide material can be guaranteed.
performed within and near the seismic area studied and not by distant
stations, which may be the only method in nuclear detection work. Ob-
servations of small earthquakes have revealed interesting facts about earth-
quake mechanisms and also given clues to the secular stress variations.
For other problems in seismology not much use can be made of very small
events. For ordinary seismological research, events of intermediate magni-
tude (around 6-7.5) have always proved to be most useful.
For an efficient improvement of the seismological observations (both
increased sensitivity and increased accuracy) a combination of the dif-
ferent procedures described will be necessary. In addition, world-wide coop-
eration is of profound significance to seismology, not only in the daily ex-
change of observations but also in the planning of future development of
station networks.
277
Chapter 9
Model Seismology
This list is far from complete. In fact, there is an equally great va-
riety of problems within model seismology as within the rest of seismology.
Japanese laboratory studies of Rayleigh waves as early as 1927 are
probably among the earliest model experiments made. The most important
development has taken place since around 1950, and nowadays model
seismology is conducted at several institutes around the world. Both
surface waves and various kinds of body waves are now studied. Earlier
models were usually three-dimensional. Such models can entail certain com-
plications from unwanted wave reflections at the different sides of the
model. Therefore, it was an important improvement when Russian and
American scientists in the beginning of the 1950's introduced two-dimen-
sional models. These consist of plates of different materials, frequently
plexiglass. If the plate thickness is one-tenth or less of the wave-length,
the plate can be considered to be two-dimensional. Important technical
improvements were made possible by the application of crystals of barium
titanate. These have the piezoelectric property of being able to oscillate in
phase with an electromotive force applied between two electrodes in the
model; in other words, the crystal acts as an electroacoustic transducer.
This substance is more efficient for this purpose than Rochelle salt, which
was used earlier in such experiments. The crystals are used both as trans-
mitters and as receivers of the waves. Recording is generally made by means
of oscilloscopes.
6.0
I(to)
6.5
7.0
7.5
5 10
5.5
I (to)
6.0
I
I
65, I
Fig. 100. Amplitude-distance curves for P- and S-waves
(horizontal components) for southeastern Europe. The ordi-
nate is/(LI)=m-iog (aIT), equation (I), Chapter 4, i. e. for
7.0 a given m the ordinate gives the variation of log (al T) with
5 10 15 20 25° distance. After J. VANtK.
Seismic Wave Propagation 283
SOURCE
;
p s
Fig. 102. p. and S-waves from a single source in a circular plate. After F. PRESS.
As wave generators PRESS used both a single source (Fig. 102) and
a dipole. A dipole source is made of two barium titanate crystals located
close to each other but oriented so as to produce opposite phases. The
amplitude diagrams and the directions of motion show good agreement
with theory both for P- and S-waves. A third experiment was also performed,
including a dipole with a fissure between the two forces, in order to simulate
nature still better. For this case the P-waves again behaved in a normal
way. whereas for the S-waves a motion was observed even in those direc-
Earthquake Mechanism 285
tions where a node was expected. The cause was considered to be a transi-
tion of Rayleigh waves, propagating along the fissure, into S-waves at
the end of the fissure-an observation which may be of importance to con-
sider also in focal mechanism studies from seismograms of natural events.
Similar results have been found in Japan. On the other hand, caution is
required in translating model results into conditions prevailing in nature.
What has been found may be representative of the model only and may not
represent nature.
In this connection, it is of interest to note that the American seis-
mologist HOWELL did not find the S-wave anomaly in his experiments
which were conducted later. Some of his results, which may be of signi-
ficance in the determination of earthquake focal mechanisms, are sum-
marized here:
I. The first motion, observed for homogeneous models, confirms the
theory for a single force, a couple and for two couples (perpendicular to
each other).
2. P-waves exhibit compression everywhere and with nearly constant
amplitude in all directions for a radial force (explosion), as expected, while
weak S-waves also appear.
3. In a homogeneous model, an un symmetric couple may give rise to
P-waves which are similar to the distribution from a radial source, but
S-waves are able to discriminate. Such findings are of importance in con-
nection with detection of explosions (Chapter 11).
4. A discontinuity may entail reduced amplitudes for P and S within
the layer which has the higher velocity. As a consequence, the first displace-
ment may become too small and may be overlooked in this medium. In turn
this may cause erroneous readings of the first motion and incorrect conclu-
sions about the focal mechanism. These results indicate the significance of
a velocity discontinuity at a distance of the order of one wave-length from
the source.
In other experiments, the wave propagation has been studied from
thermally induced cracks in a glass plate. Comparison with theory has
shown serious discrepancies, especially for the S-wave. These deviations
are ascribed to non-elastic effects, particularly at the end of the crack.
On application to real conditions in nature we have to face the question:
which approach is most suitable. theory or model experiment.
286 Model Seismology
compression, etc. While such methods are suitable for metals and glass,
they proved to be unreliable for rocks because of their inhomogeneous
structure. About 30 years ago the resonance method was introduced, in
which cylindrical specimens were used. The frequencies were observed
for different free vibrations of the specimen: longitudinal (along the cylinder
axis), transverse (perpendicular to the cylinder axis) and torsional vibra-
tions. In each case the phase velocity in the specimen can be calculated
from the length of the cylinder and the observed frequency. The free vibra-
tions are induced by means of a transducer of some kind (electromagnetic,
magnetostrictive, electrostatic or piezoelectric). The receiver can be placed at
one end of the cylinder or can be moved along the cylinder to locate nodes
and antinodes. These measurements can be performed also under high
pressure and high temperature. The method is relatively reliable even for
porous samples. The errors in the velocities amount to about ± 5 %. The
method has proved particularly suitable to the determination of transverse
wave velocities at high temperatures and high pressures.
Besides cylindrical specimens also prismatic specimens have been used,
but these require more comprehensive corrections depending upon their
shape. In a newer development of the resonance method, one observes
free vibrations of small spherical specimens-in a way corresponding to the
free vibrations of the earth observed after larger earthquakes.
In a variation of the resonance method, single crystals or crystal
aggregates are used as specimens instead of the cylindrical or prismatic
samples. When a single crystal is used as the specimen, it may be brought
into immediate contact with a piezoelectric crystal in such a way that
desired vibrations are· induced in the specimen. An alternating potential
applied to the electrodes in the piezoelectric crystal will induce vibrations
in the combined oscillator. The frequencies generally used are in the range
from 90 to 250 kc/sec, depending on the frequencies of the specimen itself.
This method can be used with small crystals of only a few mm in length.
It is also easy to apply at high temperatures, but it has not been used at
high pressures.
Pulse-transmission methods. It was not until sufficiently accurate elec-
tronic equipment had been developed during and after World War II that
this method could offer a possibility for laboratory seismology. The pulses
(50 kc/sec to 10 Mc/sec) are generated by means of a crystal transducer
288 Model Seismology
which is attached to the specimen. The travel time through the specimen
is measured either absolutely or by comparison with another specimen
(Fig. 103). A circular-cylindric shape of the specimen has proved to be
~l
E
P SPECIMEN r--
I PULSE
GENERATOR
~ ®
VARIABLE '-r-
DELAY
DUAL TRACE
OSCILLOSCOPE
Fig. 103. Principle of the pulse-transmission method. I=input signal, E=electrode, T=transducer, P=
sample, G=rubber tubing, U=output signal. After F. BIRCH (1960) and G. SIMMONS (1964). See ANDERSON
and LIEBERMANN (1966).
most suitable. Its geometrical shape and dimensions have to fulfill certain
conditions to avoid unwanted effects, such as disturbing reflections, scat-
tering of waves, etc., and thus to make it possible to reach reliable results.
Most applications of this method have concerned the direct waves,
P and S. In some experiments also transformed waves, such as PSP, have
been used. By P SP we mean a P-wave which has been partially trans-
formed into S V by reflection at the cylinder side and then by another
reflection has been transformed into P. The travel-time difference PSP-P
depends on both the P-wave and the S-wave velocities in the sample.
This method has proved useful for metals, limestone, marble. But for
more coarse-grained rocks, serious errors arise with the transformed pulse
method because of interference and reflections at the grains in the sample.
Pulse-transmission methods have been in frequent use since the be-
ginning of the 1950's. They do not require homogeneous specimens, but
can be used also for porous specimens, and moreover they can be easily
applied at high temperatures and high pressures. The accuracy of the method
is about 1-3 %.
Seismic Wave Velocities 289
For very fine-grained specimens (with grain sizes less than 0.5 mm),
a pulse-echo method has also proved to be useful. In this method a pulse
passes through the sample and is reflected back to the receiver and the travel
time is measured.
Total reflection of a P-wave at the boundary between a liquid medium
and a solid medium (the specimen) has also been applied for the deter-
mination of wave velocities (Fig. 104). When the angle of incidence i1 is
1/,
Fig. 104. Reflection and refraction at the surface of
a specimen. j = angles of incidence, v = velocities.
After G. SIMMONS (1965). See ANDERSON and
LIEBERMANN (1966).
(1)
II TIME
Fig. 105. Principle of ultrasonic-interferometry methods. T =
transducer, l=direct wave, Rl = once reflected wave, R2=twice
reflected wave. After O. L. ANDERSON and N. SOGA (1965).
Chapter 10
plans for a solution of the prediction problem. Also here, a lO-year plan
had been proposed for intensified observations in three seismic areas:
California, Nevada, and Alaska. Cost estimates for the original project
appeared very high (about 140 million dollars), and economic considerations
were a major obstacle to the realization of the original plan. On the other
hand, it has been emphasized quite correctly that this cost would correspond
only to a very small fraction of the losses in a destructive earthquake in
California. And this holds stilI true, even though later requests amount to
3 to 4 times as much. Thus, for a lO-year plan proposed in 1976 the sum has
risen to 490 million dollars, to be shared about equally between the U.S.
Geological Survey and the National Science Foundation in a broad attack
on the whole problem. In the USSR, investigations aiming at earthquake
prediction have been started in several seismic areas in Siberia, as in Garm
and Tien-Shan and on Kamchatka.
In planning for a solution of the prediction problem, we have at
present a better background than ever before, partly because of the interest
from national and international bodies, partly because of the very much
increased effectiveness in modern seismological observations and data hand-
ling techniques. It is also true to say that the time has passed when earth-
quake prediction was left only to amateurs, who underrated the inherent
difficulties, and unfortunately made the field one of bad scientific reputation.
The consequence was that for many years anyone who pretended to be
able to predict earthquakes was placed in the same category as those
who tried to construct a perpetual motion machine. Such prediction efforts
have correctly been dismissed from the scientific side. Many so-called pre-
dictions were also lacking the necessary precision. For instance, if I say that
an earthquake will occur tomorrow, I am one hundred percent right. This is
obvious from the earthquake statistics (Chapter 5). But such a 'prediction'
is completely useless. Some other viewpoints from a responsible seismologi-
cal side may also have acted as an obstruction to the development of scienti-
fically based earthquake prediction. In 1941, the famous geophysicist
GUTENBERG wrote among other things the following: 'Even if earthquake
predictions could be as accurate as weather forecasts, they probably would
do more harm than good since they would create a wrong feeling of security
and neglect of precautions agtinst eard:quake eff(cts where no prediction
has teen made and, on the other hand, they might cause unnecessary
Background and Purpose 293
earthquakes, often have caused greater damage than the earthquakes them-
selves (for example, San Francisco 1906, Tokyo 1923).
2. Earthquake prediction: through this, people could be protected
from catastrophes by evacuation and by other measures of precaution.
But evacuation with ensuing interruption of many activities is terribly
expensive, and no scientific seismologist would be willing to take that
responsibility before he had 100% certainty in his prediction. The' most
suitable procedure would probably be that seismologists for a long, long
test period make their own 'secret' predictions, and that official predic-
tions not be made until the technique has been developed to a sufficiently
high precision. Otherwise, there may again be a great risk of this whole
problem being discredited among the general pUblic.
To be of value, a prediction should tell the time of a coming shock
(preferably to the nearest day), its location (e.g. towns, villages etc., likely
to be damaged) and its intensity. From the practical point of view the
problem can be limited to damaging earthquakes, i.e. those above a maxi-
mum intensity of 7-8 in case of badly designed structures and of 9 for
specially designed structures, and the problem is naturally limited to
populated areas. From the scientific point of view prediction has signifi-
cance also for unpopUlated areas and also for deep shocks. In order to
understand the problem, it will be necessary to include also these shocks
in 'planned projects. We have also to consider that many seismic areas,
which are uninhabited today, will be built-up areas in the future. During
such expansions of the population into new areas, it is of the greatest
advantage if the earthquake engineer is able to give advice in the planning
of the building projects.
Items I and 2 represent two different aspects of the prediction prob-
lem; while I is concerned with statistical prediction, item 2 concerns de-
terministic prediction. Statistical prediction is based on past seismic history,
and can be made with fair reliability as soon as this is available. On the other
hand, deterministic prediction, i.e. prediction of isolated events, is a much
more difficult problem and for its solution a large variety of phenomena has
been investigated, as we shall learn in following sections.
There is hardly any problem that has so far-reaching consequences
as earthquake prediction. This is not only a scientific problem, but one with
the greatest concern for the general public and all its occupations. The public
The Physical and Observational Side of the Problem 295
Let us first approach the problem from the physical side and consider
the simplest possible model: a vertical string (this is essentially a one-
dimensional body) is loaded at its lower end, and the load is increased
continuously until the string breaks (for example, the load may be a vessel,
into which we are pouring mercury). If the initial load (at time t=O) is
zero, the string will break at a time 10=Fo:(dF/dt), where Fo is the load
at the breaking-point and dF/dt is the rate of increase of the load, this rate
assumed constant. Therefore, in order to predict the time 10 when the string
breaks we have to know both the rate with which the load increases with
time (dF/dt) and the properties of the string (which define Fo).
The string corresponds to the earth and the load to the stress- and
strain-generating forces in the earth. Professor H. BENIOFF in California
proposed some years ago a possible scheme for earthquake prediction which
corresponds essentially to this model. By observations of stress and strain
at an active fault and of the points in time when it breaks (i.e. when earth-
quakes occur) it would be possible to predict coming events. But accord-
ing to BENIOFF, observations would be needed for every single fault and
extending over at least two centuries, before reliable prediction could be
made. Experiences from one fault cannot simply be applied to other faults.
However, not only is the long time-scale of this project discouraging, also
the possibilities are uncertain .. The model is extremely oversimplified, since
for the real earth we have practically no reliable knowledge of the origin
296 Prediction and Artificial Release of Earthquakes
and nature of the acting forces, nor can we assume that stress and strain
vary linearly with time (rather the variation is apparently irregular, frequently
speeded up just before larger earthquakes, sometimes even reversed); further-
more, there are possible influences from external factors of various kinds
(such as activities on adjacent faults, triggering forces, etc.). The earth is
of a complicated and inhomogeneous structure, especially in earthquake
regions with numerous fractures, and we have practically no information
on breaking strength.
A direct physical approach to the prediction problem is therefore
practically impossible at present. As far as prediction is concerned, seis-
mology is certainly in a worse situation than meteorology, where it is
possible to make predictions of weather on a strict physical basis (using
large computers). In the distant future, similar predictions could be envisaged
in seismology. But there is a very long way to go before we know all para-
meters of importance and their n,umerical values for each region. There-
fore, in seismology we are forced to try indirect methods, i.e. to observe
various effects produced by increasing stress and strain, of all possible
kinds, both mechanical (tilts, slow motion, etc.) and electromagnetic. Ac-
tually, the present efforts are mainly based on 'trial and error', and in a
first attack one tries to include as many different effects as possible. To
conclude from observations of such effects something about the underly-
ing physical causes, is a very difficult interpretational problem, and has
resulted in many controversial statements, especially with regard to the
use of the various observations for prediction purposes. This difficulty is
further augmented by the fact that the observations are restricted to the
earth's surface or very near to it, i.e. we have observations only from one
boundary of the medium, from which we would need observations
throughout.
Above we have specified the earthquakes for which prediction is
wanted. In view of the facts just mentioned, we could complete that state-
ment by defining the geographical scale on which prediction is possible
or wanted, and let us distinguish between small scale, intermediate scale,
and large scale. In a small-scale project we are only concerned with a single
fault. In this case the situation is closest to the physically simple model,
and this would correspond to BENIOFF'S scheme. On the other hand, in a
large-scale (or global-scale) project we are concerned with interrelationships
Existing Projects 297
between earthquakes, first in the same earthquake belt, later maybe over
the entire globe. The projects which have been proposed by some nations
belong rather to the intermediate scale and concern regional properties.
Then, we have to bear in mind that features of single faults will be 0 bscured
as we generally have to deal with an ensemble of a large number of faults
with different characteristics. The intermediate-scale projects are certainly
very useful, but in my opinion they could be even more valuable if extended
both to small-scale investigations (of a few well-defined faults) and to the
large-scale project.
Also the time scale of any observable changes is of crucial importance
for the success of any prediction project. Changes which would be observ-
able only over geological epochs or centuries are equally useless as very
rapid changes during a few seconds or so. Intermediate-scale changes, say
of days to years, are most likely to lead to success. In general, the larger the
earthquake magnitude is, the longer is the premonitory period.
vplvs
1· 75 f-r-----\....
Electrical
Resistivity
-15%
Change
1
Geodetic Measurements
Vertical Motion-Several cm
Tilts _10- 6 ,ad.
Volumetric Strain
_10- 5_10- 6
Number of
Seismic Events
In this section I shall outline some evidence for apparent relationships be-
tween earthquakes on a large scale, a field of study which has not received
as much attention as it would deserve in the current projects. When an
earthquake occurs at some point of the earth, the stress conditions are
changed at this point. But as the whole earth constitutes one great stress
system, a change at one point would be expected to have repercussions at
all other points-even if small. This would be still more expected within
one and the same earthquake belt, which more obviously constitutes one
stress system. However, modern plate tectonics is able to offer a more
plausible explanation of this behaviour.
It has frequently been observed that seismic activity apparently wan-
ders from place to place along an earthquake belt. These wanderings are
of two kinds: oscillation and migration. See Table 23. Oscillation patterns
are found in aftershock areas and imply a more or less periodic shift of the
highest seismic activity between the two extremes of the area. It was ex-
ceptionally well developed in the Aleutian aftershock sequence in 1957,
Oscillations and Migrations 305
the oscillation period increasing with time from a few hours just after the
main earthquake and approaching about 300 days three years later. Oscil-
lations have been found not only in horizontal directions along a belt,
but also in the vertical direction as, for instance, along the sloping planes
shown in Figures 54 and 55. Migration patterns, on the other hand, are
unidirectional and only found in ordinary series of earthquakes, excluding
aftershock sequences. A repeated northward migration in Chile between
35 oS and 10 oS was found for the years 1957-60, the speed of migration
increasing from 7.6 km/day in the first cycle to 41.4 km/day in the fourth.
These cycles all began at the southern end of the belt and moved north-
wards. This scheme was interrupted by the large Chilean earthquakes in
May 1960. Similarly, a southward migration from California to Chile has
been found for 1951-8.
The east-west belt extending from the middle of the Atlantic over
the Mediterranean area and southern Asia has a tendency to show an east-
ward migration of seismic activity. Especially clear sequences occurred
in 1962 and 1976 (see Fig. 107). It should be noted that all these shocks are
significant in the sense that their magnitudes are well above the average,
except for the earthquake in Spain (August 14, 1962) which had a much smal-
ler magnitude. Note that in this sequence the magnitude increases gradually
to a maximum of 7.1 (Iran) and then decreases again. The Mediterranean
20 Bath: Introduction to Seismology
306 Prediction and Artificial Release of Earthquakes
DISTANCE KM DISTANCE KM
~ 15 'O'SPAIN
:>: o SOVIET 7·3
20 20
o ITALY 6·'
25 25
o GREECE 6·8
o CHINA 7·0
30 30
"
10 10
\ o TADZHIK 6·9
:e 1S
~ \
N
"'5!! 15
...:
OJ
z
:::>
\ "-
OJ
~ 20 SUMATRA 7·' 0 \ Vl 20
\
25 IRIAN 7·3 \0
Fig. /07. Eastward earthquake migrations along the Mediterranean to southern Asia belt in 1976 and in
1962. Distances are measured along the earthquake belts starting from the epicentre of the initial earthquake.
After each geographical name, the magnitude M is given.
and the mid-Atlantic belts join each other in the middle of the Atlantic.
Such junctions (and also bends) are places of high stress concentration and
could be assumed to be sources of migration patterns. As a matter of fact,
an oscillation pattern can be looked upon from the same viewpoint. In this
case-in an aftershock region-there are in general two 'poles' of increased
stress accumulation, at either end of the belt. It is from these poles that the
activity starts, and it becomes oscillatory with two poles, but migratory with
only one pole. A migration very similar to the one from 1962 was observed
in 1976 (Fig. 107) extending from Italy to Irian in western New Guinea.
As seen in Figure 107, the migration velocity is nearly constant and
is equal to 12 to 14 km/h. With this velocity known, we could predict the time
of occurrence of the shocks with an average error of ± 2 days, starting from
the time and location of the initial shock and assuming the locations
of all subsequent shocks known. However, the constancy of this velocity
from case to case cannot be relied upon, and, what is still worse, we have
Oscillations and Migrations 307
no knowledge of where along the belt the releases will occur. Such global
systems have to be supplemented by detailed investigations at the various
earthquake sites.
The shocks in Agadir, Morocco (February 29, 1960) and Lar, Iran
(April 24, 1960) may be looked upon in a similar way, but with more doubt:
partly because of the small magnitudes of the shocks (even though devastat-
ing), partly because of the lack of shocks in between. The migration velocity
would then be about 5 km/h.
On the basis of such indications there is reason to suspect similar
relationships between earthquakes all along the circum-Pacific belt. Com-
bined studies of time and space distribution of seismicity will certainly
prove useful. As an addition to the current projects, it is therefore of im-
portance to coordinate observations all along that belt as densely as pos-
sible, at least for a limited time. A systematic and comparative study of
deformation and energy release in aftershocks (in relation to main shocks and
foreshocks) is also of great significance.
Another related problem concerns the 'sympathetic' behaviour of
earthquakes over the globe, implying that relatively quiet intervals alternate
with intervals of higher seismic activity. It has been found that earthquakes,
even in widely different parts of the earth, frequently occur in groups,
separated by quieter periods. Statistical tests have suggested that the pheno-
menon is not real, but still it cannot be excluded that there is some real
phenomenon behind this. On the other hand, in the oscillation and migration
patterns there is implicitly contained an opposite tendency in neighbouring
areas: when activity in one of them increases, it decreases in the next one.
Also, when aftershocks following large earthquakes take place, there is
apparently a decrease in the seismic activity over the rest of the earth. It
would seemingly be easy to dismiss this as just due to a concentration of the
attention on the aftershocks (disregarding other shocks) or that the records
are so overcrowded by aftershocks that no others are discovered. Neither is a
true statement (the latter only for 2-3 hours after the main shock) and
therefore this may be attributed to some real phenomenon-but what?
It should be added that there are notable exceptions to this as to most other
observations in the prediction field.
Even though observations like these have been made for many years,
it was not until the advent of the new global tectonics or plate tectonics
308 Prediction and Artificial Release of Earthquakes
relations have been obtained. The forces produced by body tides are about
10 times as great and have therefore attracted more attention. Whereas
several scientists deny any correlation, usually on statistical grounds, others
believe in some tidal triggering of some large earthquakes. Figure 108
CRITICAL STRESS
action on the earth. Besides regular tidal triggering, there does not seem to
be any reason for unusually many or large earthquakes, as testified by a
number of scientists. The tidal action of planets is anyway quite small;
also on an earlier similar planetary constellation, no exceptional earthquake
activity took place.
Besides the triggers mentioned, which are of natural origin, there
are artificial triggers, created by human activities, which have become of
great 'significance in recent years. Since around 1940 it has been observed
at a number of places around the world that the local seismic activity has
increased some time after the construction of a dam and the impounding
of a reservoir. This phenomenon has usually been ascribed to the increased
load on the earth's surface due to the dam and especially the artificial
lake. Later observations indicate, however, that other effects also may be
involved. The observation that large underground explosions release strain
and are thus followed by a local sequence of 'aftershocks' is much more
recent. The reason is that it is only in the last few years, since 1965 and
with greater intensity from 1968 onwards, that large underground explosions
have been made (Chapter 11). Mining operations are still another trigger,
releasing small local shocks, called rockbursts (see further Section 10.8).
tl~~7T-"
~~) l)!6FV ~/
TIME (I)
d)
~(f)~ 1
1 -
1- ~1~6(fN)
/1----"------' IN
. ~t) VV~/
TIME (I) ---
e)
IN (in Fig.109,fN is identified with the release level). After this, the phenom-
enon is repeated in a similar manner and this continues over geological
epochs.
Obviously, this process and especially the condition F(t)?EIN can
be influenced, if any of the factors entering this relation can be changed
due to some external effects. We consider two cases.
1. In Figure 109c we assume that at a certain moment F receives
a constant increment of, while I and N are kept constant. As we easily see
from the figure, the effect is that the value IN is reached earlier. This means
312 Prediction and Artificial Release of Earthquakes
that the next earthquake will be released earlier than in the undisturbed
case. On the other hand, its magnitude (corresponding to the frictional
force IN) will be unaffected as well as the frequency of shocks thereafter.
2. If instead we change the frictional force fN, keeping F(t) unaf-
fected, we get the situation shown in Figure 109d. If fN is decreased by
fJ(IN), then the next shock will again occur earlier, but its magnitude
will be lower as IN is lower. Figure 109d also shows that the earthquake
frequency will be increased. The total energy released during a longer
time span will nevertheless be unchanged as long as the generating factor F
is unaffected.
By combination of Figures 109c and I09d it is easy to deduce the
development also under other conditions. This model is naturally a strong
simplification of reality, but still it can serve well as a first approximation.
In nature, F(t) does often not vary linearly with time, but shows a much
more complicated time variation with both accelerations and retardations
depending on various influences. In nature, we have also to assume that,
for some interval before the release, F(t»fN by a finite amount, corre-
sponding to the unstable area above the IN-level in Figure 109, and that
the release level is somewhat abovefN (cf. Fig. 108). A more exact treat-
ment must also take into account thatfis lower when in motion than when
at rest. Finally, in nature we have not to deal with just one fracture but
generally with a system of fractures, which can be more or less parallel.
The different fractures behave differently, especially with regard to their
values of f and N, and they can influence each other. All this leads to a
complicated time function.
Figure 10ge gives a somewhat more realistic picture with an almost
vertical fracture below a dam. In this case the load of the dam means an
increase of the force F (as in Fig. 109c). The slope of the fault surface is
of great significance, and we easily see that if this surface were horizontal
the effect of the water load would only be an increase of the normal pressure
N. Then, the water load would impede the release. For a sloping fault
surface the water load affects both F and N, and these effects can be of
opposite sign. If instead we had unloaded the surface, we could have di-
minished F (i.e. a negative fJF in Fig. 109c), which would also have had an
impeding effect on the seismic activity.
As is clear from Figure 109d, the process can be influenced not
Dams and Earthquakes 313
activity in an area, where no shocks had been noted since 1882. For in-
stance, from April 1962 to November 1965, more than 700 earthquakes
were recorded in the area, the greatest magnitude M being 4.3. As a con-
sequence it was decided at the beginning of 1966 to give up the water
injection. The area appears to be underlain by steeply dipping faults. Later,
the same results have been obtained through special water-injection experi-
ments at Rangely in Colorado. Similar effects are probably of significance
also in dam reservoirs, caused by water penetrating through cracks and
fractures in the crust.
The cases reported exclude any doubt about the seismic effects of dam
reservoirs. Cases with increased activity have naturally attracted most at-
tention, whereas the opposite case (decreased activity) can also be expected
due to a dam. One difficulty inherent in most observations of this kind is
that seismic recording at dam sites has usually been started only at or
some time after the construction of the dam. In order to make reliable
and quantitative conclusions on the role of the dam as a seismic stimulator,
we would need a long series of observations at the dam site before the
dam existed. Teleseismic observation is too insensitive to small local shocks
to act as a useful substitute. Special sensitive seismograph stations in the
near vicinity of the dam site are necessary. This is certainly something
to bear in mind in future dam construction. In the cases reported above
the information from the time prior to the dam is for the most part in-
complete, but in all cases the general opinion is that the increased seismic
activity occurred well correlated to the dam construction.
In deciding upon the causative relations between reservoir loading
and shock activity, the correlations between time series of these two para-
meters are most informative. This is especially true when the reservoir
load has been changed several times and a significant correlation with
the local seismic activity is obtained. Such diagrams, usually convincing.
have been produced in most cases reported above.
Dams are usually built in narrow passages in order to create a lake
from a river, flowing through the passage. But such gorges are only found
in complicated topography, which - at least in seismic areas - may have
originated through some earlier seismic activity. This gives us special reason
to be careful in selecting a site for a new dam, and every dam construction,
especially in seismic areas, must be preceded by careful geophysical and
Explosions and Earthquakes 3 I7
Nuclear explosions will be dealt with in the next chapter. Table 30 presents
the largest detonated underground explosions.
The Amchitka tests of October 2, 1969, and November 6, 1971,
which were announced beforehand, caused some strong protests from Japan
318 Prediction and Artificial Release of Earthquakes
Nevada:
1966 Dec 20 Dec 20-23 12 U.S. Coast and
Geodetic Survey
1968 Apr 26 Apr 26- U.S. Coast and
May 4 13 1.2 Geodetic Survey
Dec 19 Dec 19-31 33 1.3 30 aftershocks U.S. Coast and
Dec 19-23; in Geodetic Survey
all, about 10 000
1969 Sep 16 Sep 16-20 6 1.2 115 aftershocks U.S. Coast and
within the first Geodetic Survey
3 hours, with
.<1m= 1.3
Novaya Zemlya:
1973 Oct 27 Oct 27 5 2.6 Uppsala Seismo-
logical Bulletin
1974 Nov 2 Nov 2 2 (2.8) Uppsala Seismo-
logical Bulletin
Explosions and Earthquakes 319
z
o
>-
<lU
::EVl
"'<
Ow
u....,
lUW
0'"
M1 - M2 1. D
M1 - MO 2.4
DAM
TI ME
shocks).
Quite generally, any change in the natural stress conditions in the earth's
interior could lead to release. Examples are excavations within the crust,
as in mines, which may cause rockbursts. Another example is the extraction
of oil, especially in such seismically active areas as Iran with prestressed
crustal layers. Whereas relatively little is known so far about the latter
effect, rock bursts in mining areas have been known for many years. The
existing literature on rockbursts from several different countries is con-
centrated on various trigger effects and on methods for statistical prediction
of rock bursts.
Rockbursts are a daily experience in many mining areas, especially
near coal mines in central Europe or gold mines in South Africa, and they
324 Prediction and Artificial Release of Earthquakes
are also reported from North America. As distinct from this, the Swedish
iron ore mines have apparently experienced notable rockbursts only in the
last few years. This is the case with Kiruna and Malmberget in north Sweden
and with Dannemora, Gdingesberg and other mines in central Sweden.
It is known from direct measurements that the horizontal stresses
in rock generally increase rapidly with depth already in the uppermost
kilometer of the crust (Section 10.6). The horizontal stresses reach amounts
many times the weight ofthe overburden. Excavations in mining operations
lead to disturbances in the natural stress field in surrounding rock masses,
and release may take place at certain points, where the stresses exceed the
strength of the rock material. This is demonstrated schematically in Figure
Ill. In modern mining operations, excavations are done at a generally
much higher speed than earlier, with the consequence that the time variation
of the stress field will be more rapid. The natural ability of the crust to adjust
itself slowly to changing stress may then no longer be enough to avoid
sudden breakage. In addition, mining has gradually rezched greater depths
with larger pressures. These are factors of significance in explaining why
aJ .,
CRACK
:A..
CRACK
bJ
cJ
rock bursts seem to be more general in recent years than earlier in a country
like Sweden.
On the basis of recordings at the Swedish seismograph network
(Table 2 and Fig. 18), a rockburst sequence at the Grangesberg iron ore
mines in central Sweden (60. ION, 15.00 E) has been investigated for the in-
terval from August 1974 to July 1976 with 340 events in all (BATH and
WAHLSTROM, 1976). This is the first time that Sweden has produced a sequence
of shocks, analogous to an aftershock sequence, that has permitted a detail-
ed study. The magnitudes range from ML =0.5 to 3.2, with a complete cover-
age from ML~ 1.0 upwards. Tentatively applying equation (7) in Section
4.3 to the largest event (August 30, 1974) with ML =3.2, we find the seismic
moment Mo=3.5. 10 20 cgs. This would correspond to an average dislocation
of no less than 10 em over an area of 100 X 100 m. However, application of
equation (7) may not be very trustworthy in this case, as the equation was
derived for a totally different region. An average dislocation of I cm over the
seme area corresponds to ML =2.6, which appears more reasonable. The
focal depths of the events, derived from amplitude ratios of Rg to SgJ
waves, exhibit a certain variation, with the first and largest event (August
30, 1974) at greatest depth (estimated as about 1 to 2 km), rising to much
shallower depth during the summer of 1975 (probably around 200 m),
after which the depths tend to increase again. The last phase a coupled with
a decreasing rate of seismic energy release, resulting from a decrease of the
active volume of rock.
Our curves for energy and strain accumulation and release have shapes
similar to those of regular aftershock sequences, starting with a relatively
mild phase, followed by a much more active period, finally again decreasing
in activity (cf. Section 6.4). The accumulation and release are studied in
relation to various triggering factors. Mining operations may influence the
large-scale development, whereas rainfall seems to correlate well even in
details. Semidiurnal lunar tidal action is found, and certain correlation
appears between the fortnightly lunar tide and the occurrence of larger
events. From the practical point of view, rockbursts are a significant
factor in the planning and performance of future mining excavations. For
this purpose, several mines (e.g. Griingesberg and Malmberget in Sweden)
operate their own local geophone networks around the mines, to have the
development under control.
326 Prediction and Artificial Relc,bc of Earthquakes
The moral to be drawn from this and the preceding sections is that
human impacts on the solid earth have now reached such dimensions as to
fully require a much more detailed scrutiny than hitherto of the geophysical
and geological properties of areas for planned works. This is especially
true for areas which exhibit seismic activity, now or in the past. On the
other hand, much exaggerated fear of the effects has been expressed, usually
emanating from lack of knowledge of the conditions.
327
Chapter 11
?1
0)
I
.J
30
o
20 K
o
p
tJ.
10 L-~ __~__- L_ _~_ _~_ _L-~
137 138 139 140 141 142 143 144'
50 .----r--~----,----,----n
b)
40
t
sec
(.18minl
30
20
Fig. II2. Readings of PKP-waves: a) From nuclear
10 I 0 0
explosions in the Marshall Islands (P = Pretoria,
K=Kimberley. T=Tamanrasset). b) from a deep-
l
focus earthquake in South America. I=travel time,
L1 =epicentrai distance. After K. E. BULLEN and
o I
T. N. BURKE-GAFFNEY (1958).
138 140 142 144 146 148'
LEN proposed four such explosions: one in the USA, one in the USSR,
one near Australia and one in the Pacific Ocean area. The plan included
also the requirement that satisfactory precaution should be taken against
radioactive fallout. The proposal stimulated a very lively discussion with
arguments both for and against the idea. The plan was never carried out and
presently the prospects of its realization are probably more distant than
ever before.
Although explosion records would be able to give seismology large
possibilities, it is not for that reason that public interest has been focussed
on nuclear explosions. Instead this is related to the political and military
aspects of the problem. As early as 1958, a conference of experts was held
in Geneva to explore the possibilities to detect and to identify nuclear
tests. One of the proposals from this conference concerned installation
of so-called array stations (Chapter 8), also called Geneva-type stations.
Even at an early stage, seismology proved to offer one of the most
efficient methods for discovery of large explosions. From the political and
military sides many questions were addressed to the seismologists, and in
many cases sufficiently accurate answers were not immediately available.
As a consequence, responsible authorities in several countries, especially in
the USA, found it necessary to give strong economic support to seismo-
logical research on a broad basis. Since around 1960, the Advanced Research
Projects Agency in Washington, D.C., has been in charge of the American
project in this connection, called VELA, as a common name for all detec-
tion research. Within the frame of the VELA project, many institutions
have been conducting basic seismological research, both within the USA
and abroad. Even though the ultimate goal of this project is to improve
detection possibilities, it was obvious that only considerably enhanced basic
research in seismology could lead to this goal.
Besides their seismological and their military-political interests, nuclear
explosions have also another, peaceful interest, i.e. to perform large works,
as in connection with river regulation, building of canals, etc. Technically
such a use of large explosions is very well possible and has also been tested,
both in the USSR and in the USA. However, the permission to conduct
such tests is also connected with the disarmament discussions in Geneva.
Construction of nuclear power plants is another peaceful application of
nuclear energy.
330 Nuclear Tests and Other Explosions
E.Pacific Ocean
o Nevada
Marshall Islands
Semipalatinsk
Sahara
-1
-2
-3
30 40 50 60 70 80 90·
Fig. 113. Travel-time differences (at) between explosion data and JEFFREYS-BuLLEN tables for P-waves. After
ESSA Symposium on Earthquake Prediction, 1966.
Seismological Results 331
b)
....... .-
::£ -:!Y'(
~:
.A
c) ,.,.,
,~. \ I ' .
'NpN"IfJM~~'
'~ ..... ..
..1.
'
• .11••
I
~
• 1.
...
:
.•k I \
~~
d) "1~t~
It~
\
~\ .~
J. .,
.~ \ .
I
I
Seismological Results 333
but the increased precision may not be attained. If exact source data could
be published for all test explosions, this would undoubtedly stimulate seis-
mologists to further studies of their records. And these in turn would
increase our knowledge of the earth. Primarily, data on the exact posi-
tion of the source and the explosion time are needed and for atmospheric
tests also their height above the earth's surface. Even if the yield is not
published (this is apparently the most secret data), the information men-
tioned would be of the greatest scientific value. Seismological studies,
thus made possible, would certainly also contribute to providing a bet-
ter foundation for the disarmament discussions in Geneva. Even though
explosions provide seismic sources of high accuracy, it should be em-
phasized that seismologists would not suggest their application in cases
which would be hazardous to mankind, from any point of view whatsoever.
pi OS ion and the hydrophone are located on or near the SaFAR channel
(SaFAR =sound fixing and ranging), where the velocity has a minimum.
Seismological methods are also very useful; in underwater explosions,
the coupling between the explosive charge and the solid earth is as a rule
better than for explosions in other media.
3. The atmosphere, i.e. atmospheric explosions. These produce an
atmospheric pressure wave and hence are best observed by means of mi-
crobarographs. But seismological methods are also of great significance,
even though only a very small fraction of the total explosive energy is
transmitted into seismic wave energy. Atmospheric explosions have been
detonated both on or near the surface as well as at altitudes of a few kilo-
metres.
4. The ionosphere, i.e. explosions at an altitude of a few hundred
kilometres above the earth's surface. These are best recorded by suitable
magnetic instruments. Figure 116 shows such a case, obtained by an instru-
ment for recording short-period variations of the earth's magnetic field.
The magnetic method is very sensitive, and the reaction is practically in-
stantaneous. On the other hand, this method-at least as it operated at
Uppsala-proved to be insensitive to explosions in the lower atmosphere-
even to the very strongest of them. Magnetic methods are apparently among
the only ones useful for ionospheric explosions. No seismic effects are ob-
served for such explosions.
The detection methods outlined here are augmented by a number
of other observations, for instance of radioactive fallout, especially from
atmospheric tests but also from leaking underground explosions. Direct
observations by means of satellites also playa great role.
In the partial test-ban treaty of 1963, all nuclear explosions were
forbidden except the underground ones. It was then considered that suf-
Detection: Position, Depth, Origin Time 337
--
~-------
I .
Fig. 116. An explosion of 1.4 megaton at an altitude of about 400 km above Johnston Island in the Pacific
on July 9, 1962, as recorded by a magnetic instrument at Uppsala (.<1= 103'1/2).
ficiently reliable control methods were available for all tests except for
those made underground. For this reason efforts in detection work have
since then been concentrated on underground explosions and primarily on
seismological methods. While the treaty mentioned was signed by the USSR,
the USA and Great Britain, to name a few, it has not been signed by among
others France and China. The latter two have also conducted atmospheric
nuclear tests after 1963.
The fact that instrumental seismological observations still cover too short
a time interval to guarantee against surprises (see Chapter 5), has an im-
portant application to the identification of unknown events (whether ex-
plosions or earthquakes). If an event is recorded, which can be localized
to a suspected area, e.g. a known test area, and in which no seismic activity
is known, this is certainly a reason to investigate more carefully the nature
of the source, if possible. But the very circumstance that earthquakes occur
22 Bath: Introduction to Seismology
338 Nuclear Tests and Other Explosions
from time to time in areas, where they are hitherto unknown, is a warn-
ing against too hastily drawn conclusions. On the other hand, it is naturally
much easier to 'hide' an underground test by performing it within a seis-
mic area.
An increased precision in the calculation of epicentrallocations could
contribute to an improvement of the possibilities of deciding the nature
of a recorded event. To wit, it is highly improbable that an underground
explosion be set off under the ocean bottom or very near the frontier of
another country.
Seismicity studies have clearly been considerably stimulated because
of these new aspects. For instance, the seismic conditions in the Soviet
Union have attracted great interest. The Soviet Academy of Sciences in
Moscow has also contributed efficiently to these studies by publishing very
comprehensive tables and maps regarding the seismic conditions in the
USSR for the period 1911-57, a classical work in its field.
The problem of calculating the depth of a seismic event is of great
importance for the identification of the nature of the source. Even though
the precision of depth determinations is sufficiently great to establish struc-
tures as shown in Figures 54 and 55, seismic depth calculations often suffer
from errors of ±25 km. To date, the deepest underground explosions have
been made at depths of 1-2 km. If depth calculations could be made with
higher precision, events with calculated depths of say 5-10 km or more
could be referred to the earthquake category, and closer investigations
could be limited to more superficial phenomena. From this point of view
it is obvious that increased precision of depth calculations appears as one
of the most important contributions to the supervision of a test-ban treaty
for underground explosions. For this to be accomplished, increased preci-
sion of seismic measurements will be necessary, comprising for one thing
reliable timing to 0.1 sec or better, and for another thing applications of
time corrections because of local structural deviations both for the source
area and for every single seismograph station. A comprehensive investiga-
tion would be needed for the calculation of these corrections. The most
reliable depth calculations are based on the arrival-time difference pP-P.
In Chapter 4, we have seen how this method can be applied with higher
accuracy by simultaneous inspection of records from a seismograph net-
work. Spectral methods offer even more powerful techniques, applying the
Detection: Source Mechanism and Seismic Waves 339
the event had occurred exactly on a full hour, a usual explosion time at
Semipalatinsk.
In the case mentioned the considerations described were sufficient
to decide the nature of the phenomenon. Similar arguments are usually
effective in most cases when we have to deal with longer series of tests
performed within a smaller area. In the case of single tests, especially
those of smaller strength, this line of argument is frequently not enough.
Then recourse has to be taken to more detailed studies of the seismic records,
including comparison between records of known origin, both explosions
and earthquakes. Even after such a careful study, it is not always possible
to reach one hundred percent reliability.
The problem of using different properties of the records in order to
decide the nature ofthe source can in general be illustrated as in Figure 117.
If some property proves to be completely different for earthquakes and
explosions (Fig. 117a), the discrimination is one hundred percent reliable
,I
i
w
~ EXPLOSION EARTHQUAKE
~ 0)
~ r-~~~--~'-J-~-----
~ b)
their seismic effect. Therefore, their records are more similar to those of
underground tests.
There are also a number of discrepancies between records of atmo-
spheric explosions and of earthquakes at shallow depth. Surface-wave am-
plitudes are more important in relation to body-wave amplitudes for atmo-
spheric explosions than for earthquakes. A striking comparison which il-
lustrates this fact can be seen in Figure 118. In addition, records of atmo-
spheric explosions are 'smoother', i.~. they are lacking much of the higher-
frequency motion of an earthquake record. Detailed frequency analysis of
different waves may provide powerful means to discriminate between dif-
ferent phenomena. Moreover, atmospheric explosions exhibit only insigni-
ficant motion perpendicular to the plane of propagation as expected. Almost
all the motion on the seismogram is contained in the plane of propagation,
i.e. it consists of P, the SV-component of S and of R. However, some
poorly developed transverse motions are found, but these are considerably
weaker than the motion along the plane of propagation. The atmospheric
explosions-for instance, those performed at Novaya Zemlya-did not show
any traces of SH or of Love waves at Swedish stations; on the other hand,
we have found traces of higher-mode Love waves. As a consequence of
inhomogeneities in the earth's crust, transverse motion is more common
in records of underground explosions than of atmospheric tests. Clear
Love waves of fundamental mode were thus recorded from the underground
test at Novaya Zemlya on October 27, 1966 (Fig. 114) as well as from the
underground test in Nevada on December 20, 1966, just to mention a couple
of examples.
In addition to frequency analysis (spectral analysis) of recorded waves,
a detailed study of wave forms (in the time domain) may be of great assist-
ance. It has been found that short-period records of P-waves from under-
ground explosions consistently show a first displacement (compression) fol-
lowed by a second swing (dilatation) about two to three times as large as
the first one (Fig. 114). If a record is found with this typical appearance,
this feature may indicate an explosive origin. But this indication alone is
rather uncertain, as many earthquakes are able to give rise to exactly the
same wave form. In order to utilize the wave forms better, simultaneous
access to as many records as possible at widely different directions from
the source is necessary, which has already been emphasized above. On
344 Nuclear Tests and Other Explosions
the other hand, it can be maintained that a wave form, which deviates
significantly from the one described, suggests that the origin has been an
e(lrthnll!>kp !>nrl nnt !>n P¥nln,,;nn
a) 1 min
~
Skopje
j, II
b) ~
1min
:II
......'~.~~~~",.,..,..."...~'~.~1/1'10." ..~¥'/,._
Table 26. Generation of long-period surface waves (T>5 sec) from earthquakes and
explosions ' .
Table 27. Some properties which can be used for source identification'.
question is often raised as to how far from a source of a given yield it is pos-
sible to get a seismic record. A unique answer can hardly be given to this
question, as the same factors are also of importance here. The most important
of these are the following:
1. Yield. Figure 98 shows a relation between maximum recording
distance (for the most sensitive seismograph stations, excluding array sta-
tions) and yield. It demonstrates that, for instance, an underground explo-
sion of 1 kiloton could be recorded to distances of 3000-4000 km under
favourable circumstances. For surface explosions, about 1000 times as large
a yield is required for the same recording range.
2. Depth of the explosion below the surface. The deeper the explosion
is emplaced, the better is the conversion into seismic energy. That is at
least true for very shallow depths, until a depth of complete coverage is
reached.
3. Nature of the medium in which the explosion takes place. From
the USA, experience has been gained on explosions in various media, such
as salt, volcanic tuff, granite. Of these, granite and salt are the most favour-
able for the transfer of explosive energy into seismic energy.
4. Degree of coupling between the charge and adjacent medium.
Especially in the USA, comprehensive investigations, both theoretical and
experimental, have been made concerning decoupling of an explosive charge.
These indicate that it would be possible in this way to lower the seismic
output several hundred times.
A few examples will be given which demonstrate in a very striking
way the enormous variation there is in the possibility of recording an under-
ground explosion. The so-called Gnome test on December 10, 1961, in New
Mexico, USA, which corresponded to 3.5 kilotons and was made at a depth
of 800-900 m in a salt layer, was very well recorded in Sweden. On the
other hand, the Sedan explosion on July 6, 1962, in Nevada, which amounted
to 100 kilotons and was performed at a depth of about 200 m, was not
recorded at all in Sweden. The latter test was a crater experiment: the top
of the mountain was blown off by the explosion and too little energy prop-
agated downwards to give any record at our distances.
The following equation gives an approximate relation between the
magnitude mx , the yield Y in kilotons and the seismic coupling factor l/., i.e.
the ratio of seismic energy to total explosive energy:
Magnitude and Energy 349
This formula is applicable to all kinds of explosions, not only the under-
ground ones, provided that the correct value of the coupling factor is
used. The yield Y is generally expressed either in kilotons (1 kt = 103 tons)
or in megatons (1 Mt= 106 tons), by which we mean that the total explosive
energy is the same as if that amount (kt or Mt) of TNT were used. For
calculation of seismic energy we also need to know the conversion factor
between kt and ergs: 1 kt corresponds to an energy of 4 X 1019 ergs.
A number of different calculations of seismic coupling factors have
been published, of which we present a short summary in Table 28, referring
Table 28. Seismic toupling, i.e. the fraction of the total explosion energy which is
converted into seismic wave energy, for different locations of the explosion in relation
to the earth's surface (according to American data).
to an explosion of Y=20 kt. The coupling may vary considerably from case
to case. The values in Table 28 have been determined by American scientists.
Our own research in this field fully confirms the low coupling factors for
atmospheric explosions, whereas for explosions below the surface, especially
in water, we have found reason to suspect considerably larger coupling
factors than those given in Table 28.
It is internationally agreed (in 1968) that existing station networks
are able to detect underground explosions in the northern hemisphere
down to 20 kt, if the explosion has taken place in hard rock, like granite,
and that the source can be located to within 10-40 km. From the American
side it is maintained in 1971 that underground explosions as small as 1 kt
can be detected to distances exceeding 6000 km. It is of interest to compare
this statement with my graph of 1958, presented in Figure 98. Using this
graph and equation (3) above, we have in this case: distance =6000 km=54°
350 Nuclear Tests and Other Explosions
--'---'--~-~~~~'1~1~1'~~~~~
.-.
Iranian earthquake of September I,
- . --- - .-= - -..
-
- -.
. - -. .
.-- 1962 (upper figure) and from the No-
-.--
•- . - . -
. -
---~-I-~.---,~
~~~-;:::;;-~~
-
. .---.
.--- . j
vaya Zemlya explosion of October 30,
1961 (lower figure). The records are
written by the same instrument, and
-
-..-..-
...--..- ..- ..-..-. - ,
--.---.~.~.~~:~
..- . -....
-".-
therefore the trace amplitudes are di-
rectly comparable.
nitude M, which we have discussed earlier (Chapter 4), is valid for earth-
quakes. On account of the different focal mechanism for earthquakes and
explosions, the relation m-M will be different for underground explosions.
Figure 120 illustrates how a given body-wave magnitude m corresponds
to a remarkably lower surface-wave magnitude M for explosions than it
does for earthquakes. This circumstance has sometimes been considered as
the most reliable criterion for discriminating underground explosions from
earthquakes. However, a similar difference exists between deep and shallow
earthquakes, and therefore the criterion has to be combined with some in-
formation on focal depth.
There is now a large collection of observational material, including
source information, available for underground explosions made by the USA,
but we are still lacking corresponding information regarding the Asiatic
.//
6 ~ M
•
•
EARTHQUAKES
EXPLOSIONS
5
. .-
. •~.-":
4
Y'. /
1/
••
Table 29. Number of nuclear tests up to 1976, inclusive, which have been reliably
recorded by the Swedish network (Table 2).
Explanation of notation:
A Novaya Zemlya 1 Underground
B Semipalatinsk 2 Underwater
C Caspian Sea-Ural region 3 Surface
D USA, particularly Nevada 4 Atmosphere
E Aleutian Islands
F Pacific Ocean
G Sahara
H Sinkiang
I India
Year
--"=
AI A2 A4 BI CI -------~--
DI EI F2 F3 F4 GI
---..
G3 HI H4 II Sum
1954 1
55 2
56 2 2
57 4 4
58 11 2' 14
59 0
60 1
61 2 18 1 21
62 24 4 10' 40
63 3 4
64 2 4 2 8
65 9 4 2 16
66 11 9 1 23
67 14 11 27
68 11 1 20 2" 36
69 10 4 20 38
70 8 4 21 2" 37
71 11 5 7 J2 26
72 1 11 5 4 21
73 3 7 4 9 24
74 2 13 2 5 24
75 3 8 13 24
76 2 10 9 (223 )
1 USA.
2 France.
3 Probably incomplete.
Some Statistics on Nuclear Explosions 353
Table 29 gives a summary of all nuclear tests which have been recorded
by the Swedish seismograph station network (Fig. 18) up to 1976, inclusive.
Nuclear explosions were certainly started by 1945, but it took several years
until the explosions were large enough to be recorded seismically at greater
distance and also until our own station network had been created and
equipped with instruments of high sensitivity (see Table 2). Table 29 gives
an idea of the extent of the material we have to offer on this point and
at the same time clearly demonstrates the number of tests which have
been large enough to be recorded at greater distance. Military and po-
litical aspects (for instance, the partial test-ban treaty in 1963) are also
reflected in the material. In addition, there is a large number of smaller
tests which has not been recorded by us. As a rough estimate, the total
number of our recordings (415) is hardly more than 40% of the total number
of tests performed. Our statistics (Table 29) cuts the top and is therefore
still representative of the "seismopolitical" development.
The right-hand column in Table 29 shows clearly how the interval
can be divided into four periods, of which the first three show an increasing
frequency of explosions:
1. 1954-8, before the first test ban.
2. 1960-2, dominated by the large number of atmospheric explosions.
3. 1963-70, after the partial test-ban treaty in 1963, exhibiting a
steady increase of the number of underground explosions. Each of the
three years, 1968-70, has nearly the same total number, which might suggest
that some degree of saturation has been reached, technically, economically,
politically or military.
4. 1971-6, with a constant and somewhat lower frequency of explo-
sions. Notable features from this period are the continued atmospheric
tests by France and China, besides underground tests, also that India con-
ducts its first nuclear test (1974) and that the USSR expands activity on
Novaya Zemlya to include two test areas from 1973.
Table 30 summarizes the largest underground explosions which have
been made. The list includes those explosions which have given a body-
wave magnitude (m) at the Swedish network of 6.5 and over.
Even though we have based the magnitudes on Swedish records only
and even though there are still certain problems connected with magnitude
determinations, the list can be considered as relatively homogeneous, es-
23 BAth: Introduction to Seismology
354 Nuclear Tests and Other Explosions
Table 30. The largest underground nuclear explosions (m~6.5) up to 1976, incl.
American tests 1 :
Russian tests:
1 For further detailed information on recent American tests, I refer to SPRINGER and KINNAMAN (1971, 1975).
2 The largest magnitude for any explosion so far, partly due to favourable coupling.
3 Estimated to correspond to 6 megatons by the American Atomic Energy Commission, and rated as the
strongest underground test so far performed.
Table 31. Soviet explosions within the area of the Caspian Sea and Ural Mountains
up to 1976, incl. Magnitudes are calculated as in Table 30.
Date Location Magnitude Remark
(m)
in Vienna, among other things for regulation of the water system of the
Caspian Sea. According to records at our stations, explosions partly with
this purpose have been carried out already for several years. In Table 31,
I have summarized all those explosions, known to us, which could be of
interest in this connection. Coordinates, given to the nearest full degree,
have in most cases been determined by the NElS (USCGS), whereas
magnitudes (m) are based on Swedish records. The explosions in Table 31,
which were carried out during the 1950's, were all chemical, according
to Russian reports. As we see from Table 31, the explosions during the
1960's are considerably larger, and most or all of them may be nuclear.
The relation between many of the listed explosions and the Caspian Sea
and its water system is quite obvious. In addition to these cases, there are
also others, both east and west of the Ural Mountains. The Russians have
an active programme for the application of nuclear explosions to extraction
of ores, oil and natural gas, as indicated for some of the cases listed in
Table 31.
The performance of large nuclear explosions, especially for peaceful
purposes, is no doubt a major technical achievement. But it appears to
require a far greater achievement also to have such explosions under
control politically.
Chemical Explosions 359
- 'h
"-
---
-=-
"",
-~
-~
Ii. -
Fig. 121. Submarine mine explosion at Sandhamn (Baltic Sea) on August 6, 1957, as recorded by a short-
period vertical-component Benioff seismograph at Uppsala. In order of arrival, the waves are PgI, Sgl and
Rg, the latter with the largest amplitudes.
360 Nuclear Tests and Other Explosions
that the request for direct inspection at the explosion site has been dis-
cussed. For instance, a chemical explosion would not give any radioactive
fallout.
Smaller explosions of conventional nature are frequently recorded and
they are also of interest from the seismological point of view. A few ex-
amples may be given. Figure 121 shows the record at Uppsala of the detona-
tion ofa submarine mine near Sandhamn in the Baltic Sea in August 1957.
The distance is 100 km. The amplitudes (especially of the Rg-wave) are
considerable, and a weak record was obtained as far as our station at
Kiruna. The explosion occurred under· water, which is the reason for the
very good coupling to the solid earth.
The explosions in the Kiruna iron-ore mines are regularly recorded
at the Kiruna station. It is true that to a certain extent they disturb our
records, but in general we have no difficulties in separating the explo-
sions from earthquakes in these cases. Figure 122 shows a typical record
at Kiruna of one such explosion; the distance is about 10 km. Besides
p- and S-waves (denoted Pg2 and Sg2), there is a wave about 10 sec after P
which has been reflected at the base of the earth's crust at a depth of
30-35 km.
Especially since the 1960's, the Russians have performed a long
and intensive series of large chemical explosions on the Kola Peninsula.
The charges are of the order of 50-100 tons of dynamite and records are
as a rule obtained at our stations at Kiruna, Umea and Skalstugan, in
exceptional cases all over our network. The explosions are probably con-
nected with mining activities in the area, but also construction of power
plants, river regulation, etc., are possible reasons. Other explosions are
frequently carried out in the Soviet border regions, especially in the Lake
Ladoga area and around the coasts of Esthonia.
A comparative study of our records of chemical explosions in and
near Fennoscandia and of earthquakes in the same region would no doubt
yield valuable information for discrimination purposes. Because of the short
distances involved, such research would hardly be of any value for detec-
tions by means of records at greater distance. On the other hand, such a
study would be of essential importance in seismicity studies in Fennoscandia,
considering the large number of explosions carried out in the area. On the
average, the Swedish network can locate around 50 events per month within
Scandinavia and its immediate surroundings. Of these, no less than 80%
are explosions. Under these circumstances reliable discrimination criteria
appear as very significant for seismicity studies in such areas. The following
preliminary results may be mentioned:
1. Clear Rg-waves are recorded on short-period vertical-component
seismographs up to distances around 500 km from explosions, but not from
earthquakes, unless these are very shallow, i.e. with focal depths less than
2-3 km (as rockbursts).
2. The amplitude ratio Pgl/Sgl (or between any other pair of P- and
S-waves) is generally larger for explosions than for earthquakes.
3. Combination of travel times of different waves through the crustal
layers may provide information about the focal depth, which in turn may
serve as a discriminating factor.
There are exceptions to these rules, and in several cases discrimination
may be very difficult or impossible.
Besides discrimination problems, we note also the extensive use of
362 Nuclear Tests and Other Explosions
Chapter 12
Planetary Seismology
Since the first satellite was launched by the Russians in October 1957.
development has been very rapid in the exploration of outer space. This
development is of great interest also in seismological research and in geo-
physical research in general. In this chapter, we shall consider geophysics
as a whole and not just restrict ourselves to the seismological aspects.
Furthermore, we shall concentrate our attention onto the moon, as this
is the first heavenly body that has given any direct results in this con-
nection.
Geophysics and moon research-this sounds like a contrast between.
on the one hand, a research branch which is concerned with the earth,
especially the physical properties of its interior, and, on the other hand.
research which is directed outwards from the earth to outer space. How-
ever, there is really no such contrast. Instead, geophysics can expect very
useful impulses and results from such direct planetary observations, as
have now been realized in the case of the moon. The closer investigation
of the moon and the planets means nothing else than a new epoch in the
development of geophysics-for the first time, man is provided with bodies
comparable to the earth in size for direct inspection. It may perhaps seem
as if this would not diminish but rather increase our problems. To a certain
extent this is true, but on the other hand, the new observations will provide
a better check on our theories about the earth itself. It may be easy to
formulate hypotheses, when observations are lacking. But we are not satisfied
with hypotheses alone-we want to know the truth, and then observa-
tions from other planets will be of great significance also for geophysics.
There are several examples demonstrating how increased knowl-
edge about the moon could influence and enhance our knowledge about
our own earth. Here, I shall limit myself to a few of these. The base of
the earth's crust, the MohoroviCic discontinuity, is generally considered
to be a chemical discontinuity between overlying basic and underlying
364 Planetary Seismology
Table 32. Some data for the earth, the moon, Venus and Mars.
Planet Mass Mean radius Volume Mean density Gravity at
surface
1012 megaton] km 10 9 km 3 g/cm 3 cm/sec 2
1 I megaton = 10 6 tons.
plished facts for many years-BULLEN'S papers in this field extend over
more than 25 years back in time. On the other hand, access to direct ob-
servations on or near other heavenly bodies has become a reality only in
the last decade.
Our present knowledge about the moon's internal constitution can
be compared with our knowledge about the earth's interior around the
turn of the century, before seismic recordings were available to a greater
extent. It is probably correct to say, however, that the basis is more favour-
able in the case of the moon's exploration, especially as seismological
experience from the earth can be utilized in the planning of experiments
on the moon. On the whole, we could expect a repetition of the develop-
ment of seismology during the present century-but now instead con-
cerning the moon. With all probability this development will be considerably
faster than in the case of the earth.
Guided by data as collected in Table 32 and theoretical considera-
tions, certain ideas have already been formulated regarding the interior
of the moon. A usual model for the moon's density assumes a surface
value of 3.28 g/cm 3 and a value of 3.41 g/cm3 at the centre. This is the
simplest model derived from the compression of the material under increas-
ing pressure. Then it is also assumed that the composition of the moon is
chemically homogeneous and that no phase changes occur. Also spherical
symmetry is assumed. However, other facts indicate that there are devia-
tions from hydrostatic equilibrium and that density variations exist laterally
and not only in a vertical direction. The pressure at the centre of the
moon has been calculated to be slightly less than 5 X 1010 dynes/cm 2 ,
approximately the same as the pressure at a depth of only about 150km
in the earth. These and other calculations await verification by direct
observations on the surface of the moon. Among these, seismic obser-
vations play a significant role.
Whether or not the moon has a central core, corresponding to the earth's
core, is still an unsolved problem. Sometimes, it has been suspected that a
central core is lacking. This has been based partly upon the relatively low
mean density of the moon (3.34 g/cm 3), partly upon the large rigidity of the
366 Planetary Seismology
lunar body. The latter is proved by the shape of the moon, which is elongated
in the direction to the earth. This figure has arisen in an earlier epoch as
a consequence of gravity effects and still remains. Whereas remanent mag-
netization has been found in rock samples from the moon, as well as in
recent observations in the vicinity of the moon, it appears as if the moon
lacks a present magnetic field of any significance. This is a very interesting
observation which seems to indicate the absence of a moon core and at the
same time lends support to the theory that the earth's magnetism has its ori-
gin in the fluid core. However, the question about the existence or non-exis-
tence of a moon core can probably not be fully settled until we have obtained
the corresponding seismic records from the moon.
An essential contribution to the study of the topography of the
moon surface was made by the Russians in 1959. Using a very much de-
veloped technique they succeeded for the first time, by means of Lunik III,
in taking pictures of the rear side of the moon, which up to that time had
been completely unknown to us. Among the most striking discoveries was
a moon crater about 300 km in diameter, i.e. considerably larger than
the largest crater known up to that time (about 200 km in diameter). After
the publication of the Russian moon pictures, it was advocated that the
moon had been the scene of large catastrophes-violent collisions with
other heavenly bodies which had caused lava flows over the moon's sur-
face, this in turn creating the so-called seas. At first, it was believed that the
rear side of the moon was much smoother and had fewer contours than
the front side. But further photographic work, both by Russians and Ameri-
cans, revealed that the rear side resembles the front side in all its properties.
Numerous photographs have been taken on later moon flights, especially by
the American expedition in July 1971 (Apollo 15).
Before landings on the moon were undertaken, careful investigations
were made of the surface structure. Among others, a 'moon-scraper' was
used to test the properties of the surface. It had been feared that large
parts of the moon might be covered by soft material of volcanic origin.
People and instruments could then be in danger of sinking through this
material to unknown depth. However, both the results from the moon-
scraper and the placing of heavy photographic equipment on the moon
The Moon's Magnetic Field, Topography and Origin 367
double planet or that the moon was captured from outer space by the
earth. However, Professor BULLEN in Sydney, with collaborators, has proved
that phase changes in the earth's interior with reasonable assumptions
would be able to generate sufficient energy to eject the moon from the
earth. In a way, this throws us back to the old hypothesis, even though
many problems remain to be solved.
This idea about the moon's possible origin from the earth, which
was proposed by BULLEN in 1951, seems to be relatively~ unknown but
deserves to be briefly described in this connection. According to certain
newer ideas, the earth's core inside 2900 km depth, has the same chemical
composition as the surrounding mantle (silicates), but by a phase transi-
tion under high pressure and high temperature, the core material is in a
so-called metallic phase (Chapter 7). A planet with such a core of very
small size is in an unstable state, and the stable state corresponds to the
normal (molecular) phase. The phase change to the stable state entails
liberation of energy, which under certain circumstances could be enough
to eject the moon from the earth. The Indian scientist DATTA showed in
1954 that this is possible quantitatively, if we start from an original earth-
moon with three phases, a normal phase, X, and two high-pressure phases,
Y and Z. The disappearance of the phase Z would be connected with
enough energy liberation to eject the moon from the earth, provided that
the radius of the phase Z amounted to at least 1500 km and that its density
were at least 18.5 gjcm 3 • These assumptions appear to be reasonable, but
cooperation is required with some resonance effect which could bring the
body over a certain potential barrier.
Age determinations oflunar rock samples, collected during the Apollo
landing in July 1969, gave surprisingly high values, about 3.5X 109 years.
Age determinations of samples collected during later Apollo expeditions
have yielded even higher values, up to about 4.6 X 109 years. Such results
will be considerably amplified by investigation of rock samples from other
localities on the lunar surface as well as from beneath it, through planned
drilling to about 3 m depth. The presently most-favoured idea about the
moon's origin is that it was captured by the earth about 4X 109 years ago.
Se1enophysical Phenomena: Volcanic Eruptions, Moonquakes, etc. 369
will cause heavy stresses in the moon's crust, which could provoke breaks
and displacements. Observation of radiofrequency radiation demonstrates
a temperature variation of only about llOo. This corresponds to condi-
tions at a depth of about 40 cm below the moon's surface. Likewise, the
Apollo 15 heat flow measurements reveal that the diurnal temperature
variation is almost negligible at 50 cm depth. These observations show not
only that the superficial layer consists of material with very low heat con-
ductivity, but also that the stresses produced must be restricted to the very
top leyer. Nevertheless, it has been found that a large number of small
moonquakes must be ascribed to the diurnal temperature variation on the
moon's surface. These quakes, so-called thermal moonquakes, are explained
as slumping of lunar soil triggered by diurnal thermally induced stresses.
It has been reported repeatedly that haze or fog has been observed
on the moon, especially in some crater. Most likely this consists of gas
emanating from the moon's interior and is not a result of condensation
in any thin atmosphere. Of particular interest are the much discussed
observations of presumed volcanic activity on the moon that were made
from the Crimea Observatory, for the first time in November 1958, and
since then on several other occasions. The question of the possible existence
of a very thin moon atmosphere has not yet been answered definitely. It
does not appear impossible that a thin upper atmosphere and an ionosphere
could exist, to some extent resembling conditions on the earth. A few
observations of light flashes, interpreted as due to meteorites passing
through the outer moon atmosphere, could be taken as an indication of
this. Ionosphere scientists and meteorologists will probably also have an
interest in closer investigation ofthe moon, even though some of the planets,
such as Venus or Mars, will have more to offer for these branches.
exponentially with their decreasing size, and that the whole surface of
the moon has been covered by craters over and over again. Such results
have been of significance in connection with studies of craters on the earth
as well as for a better apprehension of conditions in space, of importance
for future space flights among other things.
By means of smaller explosions, a method is offered to generate
seismic waves both regionally and through the whole of the moon, whose
records would clarify the layering in its interior, the possible existence of a
lunar core, etc. By means of records of moonquakes, the seismic conditions
on the moon could ultimately be mapped.
An interesting seismological experiment was carried out in connection
with the moon landing of Apollo 12 in November 1969. After takeoff
from the moon, the ascent stage of the lunar module was shot against the
moon's surface. This generated an artificial moonquake which gave strong
records on the seismometers placed 75.9 km from the impact point. Most
remarkable in this connection is the long duration of the record-no less
than 55 minutes-which gave rise to many discussions (Fig. 123). With
I
Ili_.'*~I.t•.t.1."11"1"'--"';LG~"""""'~==
lM Impact __ 6 nm
20 Nov. 1969 l'
22 :17:41.2 10 minutes
Fig. 123. Seismic signal received on the lunar long-period vertical-component seismometer from the Apollo
12 lunar-module impact. After G. LATHAM et al. (1970). 1 nm= 1 nanometre= 10- 9 metre = 1 millimicron.
a mass of 2383 kg and an impact velocity of 1.68 km/sec, the kinetic energy
of the lunar module corresponds to a body-wave magnitude m=5.0 at
100% conversion into seismic wave energy [using equations (1) and (2) in
Chapter 11], to m=4.6 at 10% energy conversion, and to 111=4.2 at 1%
energy conversion. However, it is estimated that the conversion factor may
be much smaller, around 0.001 %, i.e. 10- 5 of the kinetic energy. For such
a conversion we get m = 3.0. The ascent stage, shot down from Apollo 13
in the spring of 1970, had a larger mass (13 925 kg) and a higher impact
velocity (2.58 km/sec). The seismic magnitude m is calculated to be 5.4,
5.0, 4.6 and 3.5, corresponding to the four cases, respectively, just dealt
with. A 4-hour-long seismic record was then obtained at a distance of 115 km.
Later experiments during the following Apollo flights gave similar results.
For comparison, on February 4, 1967, an earthquake occurred inVarmland,
Instrumental Observations and Models of the Moon 373
meteor impacts. This means that both reasons for the moonquakes were
exterior to the moon and any interior activity, corresponding to that known
for the earth had so far not been discovered. The observation of moonquakes
due to lunar tides is interesting because it sheds light on an old problem
concerning the earth, i.e. whether earthquakes are released by tidal trig-
gering action or not. The experience from the moon seems to justify the
assumption of such an effect, even though it is generally hidden by other
more important phenomena on the earth. For the moon it did not appear
clear whether the tides act as generators of quakes, i.e. causing both strain
and its release, or if they act simply as triggers, i.e. releasing strain existing
in the moon for other reasons.
It is of considerable interest to note that moon recordings generally
do notresemble those regularly obtained on the earth, except for those due
to such phenomena as volcanic tremors, microseisltls, landslides, etc.
Distinct phases as observed in earthquake records do not show up on the
moon in the time-compressed record of Figure 123. With better resolution,
however, separate phases have been found, corresponding to P, Sand
surface waves, even though these are much less conspicuous than on the
earth. If any quake activity should exist, similar to that on the earth, then
the time of observation would have been long enough to record many of
these. At present, we can speculate either that tectonic activity is absent
on the moon, at least it is not at all comparable to the one on earth, or that
the moon's interior is of such a nature as to prevent wave propagation to
greater distances. The latter idea obviously conflicts with suggestions of
high quality.
With the installation of the seismograph equipment of Apollo 15 in
July, 1971, the first triangular array was established, further developed by the
installations of Apollo 16 and 17. This naturally enhanced the possibilities
of locating seismic events. The most remarkable finding was moonquakes at
depths of 700-1200 km with magnitudes of2 to 3. While generally no larger
moonquakes have been observed, the depth is quite remarkable. Among
other events we note swarms (Section 6.4) of moonquakes with one shock
every second hour and lasting for several days, but still with unknown
reason. It has also been possible to locate specially active zones - one of
which is responsible for about 80% of the moon's seismic activity.
With the much more abundant data which gradually became avail-
Instrumental Observations and Models of the Moon 375
able, it was possible also to conduct some statistical studies of moon quakes.
For instance, in the frequency-magnitude relation log N=a-bM, cf.
Section 5.2, it was found that b"=".2 for moonquakes as distinct from its
value around I for tectonic earthquakes. Such a high value of the b-coeffi-
cient is characteristic for volcanic earthquakes, and could suggest micro-
fracturing, possibly due to thermal stresses.
Guided by our seismological knowledge about the earth, quite a
number of theoretical calculations have been made regarding the seismic
conditions of the moon. Among such calculations, we may mention deter-
mination of preliminary travel times for P- and S-waves, assuming con-
stant velocities of 8.0 km/sec and 4.6 km/sec, respectively, for the entire
moon. On the basis of these assumptions, it will take a little more than 7 min
and 12-!- min for a P-wave and an S-wave, respectively, to pass straight
through the moon. Later, these calculations have been extended to var-
ious layered models of the moon, also to a number of body waves in addi-
tion to P and S (as pP, PP, sS, SS). See NAKAMURA and LATHAM (1970)
and DERR (1970). Other theoretical calculations concern surface waves.
Among other things, dispersion curves for Rayleigh waves with periods
between 10 and 100 sec have been drawn for two moon models with dif-
ferent layering in the moon's crust. Similarly, the free vibrations of the
moon have been calculated under the assumption of certain models. Ob-
servations of spheroidal moon vibrations are expected to be able to dis-
criminate between different moon models. Such calculations, of which
just a few examples have been mentioned here, are mostly theoretical
and speculative. But they will certainly become of great importance
by comparison with direct seismological observations of the moon.
More recently, it has become possible to place considerations of moon
models on a more reliable basis thanks to direct observations. See Figure
124 and Table 33. The velocity structure is based on travel times and
amplitudes of recorded seismic waves, especially P, together with informa-
tion from theoretical seismograms and laboratory measurements of seismic
wave velocities in rock samples from the moon. A two-layered crust with
total thickness of no less than about 65 km is suggested, overlying a litho-
sphere which extends to depths around 1200 km. The existence of remarkably
deep moonquakes, at depths of 700-1200 km, suggests that the interior is
solid enough to such depths to hold stresses. The innermost part, astheno-
376 Planetary Seismology
10
VELOCITY Vp
KMISEC
10
DEPTH
h KM
20
-25 KM
30
40
....
III
=>
'"
U
1""
50
60 "MOHO"
LITHOSPHERE
70 Fig. 124. P-wave velocity profile in the crust and uppermost
mantle of the moon. After ToKSl)z et al. (1972), modified.
sphere, extending from about 1200 km depth to the moon's centre, is prob-
ably partially molten, as indicated by stronger attenuation of S-waves and
absence of moonquakes. There is no correspondence to the earth's core,
unless one would like to look upon the inner asthenospheric globe as a
certain counterpart, even though it better corresponds to the earth's lower
mantle. Comparing the gross features of the earth's and the moon's interior,
we could say that the moon exaggerates the outer parts of the earth, but
Table 33. Model of the moon's internal structure.
suppresses the inner parts. For instance, taken in relation to the respective
radii of the moon and the earth, both the crustal thickness and the depth to
the deepest quakes are 6 to 7 times as large for the moon as for the earth.
Other geophysical-or more correctly selenophysical-measurements
should also be conducted on the moon, such as heat flow and radioactive
measurements, observations of moon tides, gravity measurements and
selenodetic measurements. Several of these are already included in the man-
ned lunar landing projects, besides the coltection of lunar rock samples.
Selenothermal measurements ought to give information about the thermal
history of the moon, and thus contribute to the solution of the problem of
the origin of the moon and our planetary system. The shape of the moon
should be accurately determined, and also an exact coordinate network
(longitudes and latitudes) will be necessary.
A recent observation on the moon concerns so-called mascons, i.e.
mass concentrations, which exhibit themselves by increased gravity. Their
origin has been much discussed, and the suggestions vary from meteoritic
origin to an origin from the moon's interior. Besides the existence of
mascons, there are several other indications of relatively low internal tem-
peratures, probably below the melting point, within the moon, e.g. results
from the Apollo surface magnetometer experiment, furthermore the shape
of the moon and the existence of very deep quakes, which both testify to
considerable internal strength. Also direct heat flow measurements on the
moon's surface give similar results. The heat flow has been measured as
30 ergs/cm 2 sec, i.e. 0.7.10- 6 cal/cm2 sec or only about half of the earth's
average value, estimated as 1.5· 10- 6 cal/cm 2 sec. Only preliminary estimates
of the moon's internal temperature have been made, all of them lying below
3000 °C and some even below 1000 0c.
Observations of cosmic radiation, not to mention astronomical ob-
servations, both optical and in the radiofrequency range, could be made
with advantage from the moon as well as all observations for which the
earth's atmosphere is a disturbing factor. Among these we may mention
all kinds of radiation measurements, such as radiation from the sun and
from the earth. These and other observations have been made with rockets,
which, however, are unable to provide continuous observations over a
longer time span. Satellites, however, provide such a possibility. For in-
stance, for observations of corpuscular radiation from the sun, distances
378 Planetary Seismology
on the earth are too short to provide sufficiently long base lines. But simul-
taneous observations from the earth and the moon would be able to furnish
important information about the propagation and distribution of this
radiation. The moon contains such a record of the sun's development as
is impossible to find in meteorites or in any substance on the earth. For
4.6 X 109 years, the moon has been exposed to an uninterrupted solar wind.
This means that particles from the sun can be found on the moon and
corresponding samples have now been collected for laboratory investiga-
tion on the earth. Also a number of observations of our own earth could
be made with advantage from the moon, e.g. of the tail of leaking atmo-
sphere that sometimes has been suggested to exist. At present, it is mainly
a question of technical facilities, whether many of these observations are
better made on the moon's surface or from a space laboratory. The low
gravity and the practically complete absence of air resistance make the
moon very suitable as a platform for continued satellite and rocket ex-
periments on our way towards outer space.
A laser reflector placed on the moon in July 1969 (Apollo 11) has
made it possible to measure the earth-moon distance to an accuracy of
about 15 cm. This increased accuracy has great significance for many studies,
such as of the moon's motion and of continental drift and polar wandering
on the earth. Therefore, a network of observing stations around the earth
would provide accurate means for testing hypotheses as expressed in the
new global tectonics (Chapter 6).
Even though interest at present is naturally focussed on the moon,
preliminary calculations, similar to the ones described above for the moon,
are being made for the planets next in line for manned landings, i.e. Venus
and Mars.
As the mass and the mean density of Venus are nearly the same
as for the earth (Table 32), these two planets have presumably almost
the same composition. Observations on Venus may therefore have a special
relevance to observations on our own earth. It is considered on good grounds
that Venus has a core, which is metallic and liquid, like the outer core of
the earth. Accurate determinations of the radius of the Venusian core
would make a discrimination possible between the two most important
hypotheses for the boundary of the core: whether this is a chemical dis-
continuity or represents a phase change (see Chapter 7). We see that this
Instrumental Observations and Models of the Moon 379
mantle boundary only 2900 km beneath our feet is less certain than our
knowledge about the outer parts of the sun.
The recent direct observations on the moon have stimulated an
enormous literature and also raised a number of new problems not envisaged
before. But there is no doubt that with these observations new pathways
are opened up to science, which will lead to a more reliable apprehension of
the universe and its development.
381
Chapter 13
defend a thesis for the doctor's degree, if he fulfills the general admission
requirements.
3. Educational Program
3.1 General
The doctor's degree program is planned so that it normally requires four
years' full-time study. The course-related work corresponds to approxi-
mately 40 points and is mainly concentrated in the first two years. This
includes both elements which are common to all students and elements
which are directed towards the scientific studies particular to the individual
dissertation.
At the beginning of the program, an individual study plan which the
student intends to follow is defined. This plan is worked out between the
student and his advisor.
3.2 Courses
3.2.1 Methodology and introduction, 8 points comprising:
a) Physical geology (features and formations of the surface of the Earth,
constitution of the crust, external and internal processes, corresponding,
for example, to HOLMES: Principles of Physical Geology).
b) Statistical processing of observations (distributions, means, correlations,
significance calculations, spectral theory, least-square method).
c) Applied mathematics (contour integration, conformal mapping, sta-
tionary-phase method, series integration, Bessel functions, Legendre
functions, integral transforms, matrix algebra, calculus of variations,
integral equations; BATH: Mathematical Aspects of Seismology).
d) Computer programming, in case this is not included in the student's
preparation (the ability to write the necessary programs may be best
acquired by auditing one of the programming courses given in other
areas).
e) Spherical trigonometry and map projections (familiarity with spherical
trigonometric formulas, especially sine and cosine relations, as well as
the most common map projections).
f) Introduction to seismology (BATH: Introduction to Seismology) and to
384 Seismological Education and Practice
3.3 Dissertation
For the doctor's degree it is also required that the candidate write and
publicly defend a scientific presentation. Specifications of the form of the
dissertation and of the public defence are found in §§ 59-69 Declaration
(1969: 50) on education in the philosophical faculties.
The dissertation shall be of such quality that it is judged to fulfil
recognized, reasonable requirements to be accepted, as a whole or in sum-
mr.ry, for publicetion in an international scientific journal of high quality.
The choice of subject for the thesis is normally made in agreement
with the advisor.
4. Instruction
The Seismolofic21 Institute provides instruction for the basic courses for
the doctor's degree to the extent that instructional steff is available. Beyond
this, lectures and seminars in specialized areas as well as guest lectures are
arranged.
5. Advisors
Candidates accepted to graduate studies are entitled to supervision by an
advisor, full-time students for four years, and part-time students with uni-
versity positions as assistants or technicians for five years. Other part-time
students are entitled to research supervision of corresponding extent.
Upon entrance into the program, each student is assigned an advisor who
is continuously available to supply guidance concerning the course of studies
and the selection of course literature, etc. Later, usually when the student
has completed some of the basic courses and begun to consider the nature
of the dissertation, the student will work with a research advisor who need
not be the advisor first assigned and in the normal case directs the special
studies and research leading to the dissertation. Normally, this research
advisor is also an examiner for the doctoral dissertation.
6. Examinations
Examinations for the doctor's degree in seismology cover the basic courses
by means of written or oral examinations, both at the terminations of
courses and at other times when required.
Those who wish to carry out studies and dissertation work without
being accepted as a doctoral candidate are entitled to undergo examinations
for the doctor's degree.
Students are entitled to transfer credit for examination passed in
graduate studies in a philosophical faculty even at another university. This
also applies to examinations in graduate studies at colleges within Sweden.
7. Miscellaneous Information
Because of the international nature of the research community, all instruc-
tion (lectures, seminars, etc.) is given as a rule in English. The students also
use English in their seminar talks, which seems to give the students practice
for participation in international scientific conferences.
13.2 Exercises
In this section we give a number of exercises, by which the reader will get
an opportunity to test his knowledge acquired from the reading of the
present book. All the problems are relatively simple and of such a nature
as a seismologist is confronted with in his daily work with records.
1. Determine the epicentre and the origin time for the following event:
Exercises 387
h m s h m s
Uppsala i 22 50 57.0 Skalstugan iPn 22 48 11.2
iSg 22 51 03.7 i 22 48 34.5
iSg 22 48 57.2
Kiruna iPg 22 48 16.9
i 22 48 30.1 Umea ePn 22 48 24
iSg 22 49 00.7 iPg 22 48 37.0
iSn 22 49 07.5
is* 22 49 22.2
iSg 22 49 33.0
Hint: Use the Jeffreys-Bullen travel-time tables and a map of the Nordic
countries, preferably of scale 1: 5 X 106 •
Answer: West coast of Norway at 66.6 ON, 13.6°E. Origin time
=22 47 22.
2. Same problem as 1 for the following event:
h m s h m s
Kiruna iPn 00 09 48 Umea iSn 00 11 28.1
iSn 00 10 42.1 is* 00 11 43.7
iSg 00 11 05.9 iSg 00 12 03.0
Skalstugan eSg 00 13 35
Answer: Northwest Russia, 68.2 ON, 31.6 °E. Origin time =00 08 37.
3. Same problem as 1 for the following event:
h m s h m s
Uppsala iPg 04 16 56 Goteborg ePn 04 16 17
iSn 04 17 34 eSg 04 16 57
iSg 04 17 53
Karlskrona iPg 04 15 40
Ska1stugan eSn 04 19 00 iSg 04 15 48
iSg 04 19 49
Answer: Southern Baltic Sea, 55.6 ON, 15.0 o E. Origin time=04 15 28.
4. Are the following phases possible and, if so, how do they propagate:
PcPPKP, PPPKP, SKSP, SKKKP, SKPSKP, PKIKKIKP, sPcP, pSKP,
SeSSeS?
5. Suppose we have a given earthquake and a given station, such
that a) both Sand SKS are recorded; b) both P and PKP are recorded.
Explain how this is possible and within what distance ranges a and b can
happen.
6. Suppose we have records in three components (E, N, Z). Which of
the following waves can be used for determination of the direction to the
epicentre: P, PP, S, PeS, ScS, SKS, L, R?
7. Given the following earthquake: November 9, 1963, 21 15 30.4
GMT, 9.0 oS, 71.5 oW, western Brazil, depth about 600 km, magnitude
m = 7.1. Calculate the following quantities for the station at Uppsala:
a) Epicentral distance to 0.10 by means of spherical trigonometry_
b) Azimuth of the epicentre as seen from the station and vice versa to 0.1 0
by means of spherical trigonometry.
c) Theoretical arrival times of P, pP, PP, SKS, S, PKKP, P' P' using both
the Jeffreys-Bullen tables and the Gutenberg-Richter tables.
d) Compare the results from c with observed arrival times at Uppsala.
e) Angle of incidence (or angle of emergence) at the focus and at the sta-
tion of P, pP, PP, S.
f) Depth of deepest penetration of P, Sand PP rays.
Hint: For the solution of a and b, knowledge of spherical trigonometry is
required (as a rule, the cosine and the sine theorems are sufficient for all
seismological purposes). For part d, the Uppsala readings are communi-
cated:
iP 21 28 01.5
ipP 21 30 06.5
iPP 21 32 18.6
iSKS 21 37 42
is 21 38 36
iPKKP 21 44 28.2
iP'P' 21 52 51.9
Exercises 389
problem under 3 above and for a more detailed account we refer to Chap-
ter 10.
7. Recording of seismic waves to establish the nature of the source.
This field comprises several parts, of which we give the most significant
examples.
a) Nuclear test detection. Detection of nuclear explosions has now
to be considered as an important branch of applied seismology. The details
of the problem have been described in Chapter 11. Research on this prob-
lem as well as practical detection are being conducted both by government
agencies and by seismological institutes in several countries.
b) Seismic detection in connection with military operations. In such
operations, it has been customary for a long time already to listen in var-
ious media, such as in the atmosphere and in water (hydrophones). More
recently, these methods have been supplemented by listening methods
applied to the ground, using seismic detectors.
c) Seismic detection of rockbursts. Rockbursts (Section 10.8) are small
ruptures which can be located seismically by recording with a number ofre-
ceivers. Such recordings are of significance as they can locate possible crack
formation in tunnels, underground chambers, mines, etc. They may there-
fore be of importance for safety measures against major ruptures.
8. Seismic-glr.ciologicel research. Every line of work, where seismic
methods are being employed, can be considered as a branch of applied
seismology. Seismic-glaciological research offers one such example. In-
vestigations have been made of the internal structure of glaciers in many
areas, by the application of seismic refraction techniques.
On the whole, it can be said that there is an increasing demand for
seismologists all over the world with a good theoretical and experimental
background. Technically highly skilled seismolog~sts are required for the
operation of seismograph station networks and of array stations as well
as for the reading of records and other handling of data. Universities need
well-qualified people to carryon teaching in seismology on an advanced
level. This together with the practical side, of which a number of examples
have been given above, tend to make the prospects for seismologists brighter
than ever before.
395
Literature Review
The intention of the following literature review, arranged according to chapter, is both
to give the main sources upon which the present book is based as well as to give sugges-
tions for further reading. On the other hand, the intention is not to give any complete
coverage of the seismological literature, which would be practically impossible with
the great abundance of published papers. The list is rather to be looked upon as a selection
of papers suitable for further reading. In the papers listed there are often extensive ref-
erences to further literature. As a considerable part of the material in this book is based
upon observations and studies at the Seismological Institute, Uppsala, there is a corre-
sponding overweight of papers from this institute. In case publications refer to more
than one chapter, they are generally listed only on the first of these chapters. Books given
already in Chapter 13 are in general not repeated here.
Chapter 1
BATH, M., Seismology in the Upper Mantle Project. Tectonophysics 1. 261-271 (1964).
BATH, M., Seismology. Acta Univ. Ups., Uppsala University 500 Years, Fac. Sci. 8.
181-191 (Math. & Physics), and 10. 207-217 (Earth & Life Sci.) (1976).
BOLT, B. A. (Ed.), Methods in Computational Physics. Vol. 13: Geophysics (Acad. Press,
New York 1973),473 pp.
CIVETTA, L., GASPARINI, P., LUONGO, G., and RAPOLLA, A. (Editors), Physical Volcanology
(Elsevier, Amsterdam 1974), 333 pp.
GUTENBERG, B. (Ed.), Handbuch der Geophysik. Vol. 4 (Borntraeger, Berlin 1932), 1202 pp.
GUTENBERG, B., Seismology. Geol. Soc. Amer., 50th Ann. Vol., 439-470 (1941).
JEFFREYS, H., and JEFFREYS, B. S. (Editors), Collected Papers of Sir Harold Jeffreys on
Geophysics and Other Sciences. Vol. 1-6 (Gordon and Breach, London 1971-7).
LOVE, A. E. H., A Treatise on the Mathematical Theory of Elasticity (Dover, New York
1944), 643 pp.
PARASNIS, D. S., Principles of Applied Geophysics (Chapman and Hall, London 1972),
214 pp.
PRESS, F., and SIEVER, R.o Earth (Freeman, San Francisco 1974),945 pp.
SCHEIDEGGER, A. E., Foundations of Geophysics (Elsevier, Amsterdam 1976), 238 pp.
SHARMA, P. V., Geophysical Methods in Geology (Elsevier, Amsterdam 1976), 428 pp.
Chapter 2
ANDERSON, J. A., and WOOD, H. 0., Description and Theory of the Torsion Seismometer,
Bull. Seism. Soc. Amer. 15, 1-72 (1925).
BATH, M., Development of Instrumental Seismology in Sweden in 1949-1958, Geofis.
pura e appl. 43, 108-130 (1959).
BENIOFF, H., A New Vertical Seismograph. Bull. Seism. Soc. Amer. 22, 155-169 (1932).
BENIOFF, H., A Linear Strain Seismometer. Bull. Seism. Soc. Amer. 25, 283-309 (1935).
BENIOFF, H., Earthquake Seismographs and Associated Instruments, Advances in Geophys.
2,219-275 (1955).
396 Literature Review
BENIOFF, H., Long-Period Seismographs, Bull. Seism. Soc. Amer. 50, 1-13 (1960).
BORG, H., and BATH, M., The Uppsala Seismograph Array Station, Pure and Appl.
Geophys. 89, 19-31 (1971).
COULOMB, J., and GRENET, G., Nouveaux principes de construction des seismographes
electromagnl!tiques. Ann. Phys., Ser. 11,3, 321-369 (1935).
DE BREMAECKER, J. CL., DONOHO, P., and MICHEL, J. G., A Direct Digitizing Seismograph,
Bull. Seism. Soc. Amer. 52, 661-672 (1962).
GUTENBERG, B. (Ed.), Handbuch der Geophysik, Vol. 4 (Bomtraeger, Berlin 1932),
1202 pp.
HAST, N., The Measurement of Rock Pressure in Mines. Swedish Geological Survey,
Yearbook 52, 183 pp. (1958).
HAST, N., The State of Stresses in the Upper Part of the Earth's Crust, Engineering Geology
2,5-17 (1967).
WILLMORE, P., The Detection of Earth Movements. Methods and Techniques in Geo-
physics (S. K. RUNCORN, Ed.), Interscience, J, 230-276 (1960).
Chapter 3
ANSELL, J. H., Observation of the Frequency-Dependent Amplitude Variation with Distance
of P waves from 87° to 119°. Pure and Appl. Geophys. 112, 683-700 (1974).
BARBER, N. F., Fourier Methods in Geophysics. Methods and Techniques in. Geophysics
(S. K. RUNCORN, Ed.), Interscience, 2, 123-204 (1966).
BATH, M., An Investigation of the Uppsala Microseisms (Almqvist & Wiksell, Uppsala
1949), 168 pp.
BATH, M., Comparison of Microseisms in Greenland, Iceland, and Scandinavia, Tellus 5,
109-134 (1953).
BATH, M., A Study of T Phases Recorded at the Kiruna Seismograph Station, Tellus 6,
63-72 (1954).
BATH, M., Ultra-long-period Motions from the Alaska Earthquake of July 10, 1958, Geofis.
pura e appl. 41, 91-100 (1958).
BATH, M., Channel Waves in the Earth's Continental Crust, Scientia 56, 8 pp. (1962).
BATH, M., Propagation of Sn and Pn to Teleseismic Distances, Pure and Appl. Geophys.
64, 19-30 (1966).
BATH, M., An Earthquake with Exceptionally Strong Higher-mode Surface Waves, Pure
and Appl. Geophys. 66, 16-24 (1967).
BATH, M., Observations of Teleseismic Pn Phases, Pure and Appl. Geophys. 66, 30-36
(1967).
BATH, M., Average Crustal Structure of Sweden. Pure and Appl. Geophys. 88. 75-91
(1971).
BATH, M., Short-Period Rayleigh Waves from Near-Surface Events. Phys. Earth Planet.
Interiors 10. 369-376 (1975).
BATH, M., and LOPEZ ARROYO, A., Attenuation and Dispersion of G Wm'es, J. Geophys.
Res. 67, 1933-1942 (1962).
BATH, M., and LOPEZ ARROYO, A., Pa and Sa Waves and the Upper Mantle, Geofis.
pura e appl. 56, 67-92 (1963).
BATH, M., and CRAMPIN, S., Higher Modes of Seismic Surface Waves - Relations to
Channel Waves, Geophys. J. 9, 309-321 (1965).
BATH, M., and SHAHIDI, M., T-phases from Atlantic Earthquakes, Pure and Appl. Geo-
phys., 92, 74-114 (1971).
Literature Review 397
Chapter 5
BATH, M., Seismicity of Fennoscandia and Related Problems, Ger!. Beitr. Geophys. 63,
173-208 (1954).
BATH, M., Earthquakes, Large, Destructive, Int. Dict. Geophys. 1, 417-424 (1967).
BATH, M., and DUDA, S. J., Strain Release in Relation to Focal Depth, Geofis. pura e
app!. 56, 93-100 (1963).
BENIOFf, H., Orogenesis and Deep Crustal Structure - Additional Evidence from Seis-
mology, Bull. Geo!. Soc. Amer. 65, 385-400 (1954).
BENIOFF, H., Seismic Evidence for Crustal Structure and Tectonic Activity, Geo!. Soc.
Amer., Spec. Paper, 62, 61-74 (1955).
DUDA, S. J., Secular Seismic Energy Release in the Circum-Pacific Belt, Tectonophysics
2, 409-452 (1965).
DUDA, S. J., Regional Seismicity and Seismic Wave Propagation from Records at the
Tonto Forest Seismological Observatory, Payson, Arizona, Ann. Geofis.18, 365-397
(1965).
HEEZEN, B. C., The Rift in the Ocean Floor, Sci. Amer., 1-15 (Oct. 1960).
HEEZEN, B. c., and EWING, M., The Mid-Oceanic Ridge and its Extension Through
the Arctic Basin, Geology of the Arctic, 622-642 (1961).
OLIVER, J., RYALL, A., BRUNE, J. N., and SLEMMONS, D. B., Microearthquake Activity
Recorded by Portable Seismographs of High Sensitivity, Bull. Seism. Soc .. Amer.
56, 899-924 (1966).
ROTHE, J. P., The Seismicity of the Earth 1953-1965, UNESCO, Earth Sciences 1,
336 pp. (1969).
SCHEIDEGGER, A. E., Physical Aspects of Natural Catastrophes (Elsevier, Amsterdam
1975), 289 pp.
SCHNEIDER, G., Erdbeben, Entstehung, Ausbreitung, Wirkung (Enke, Stuttgart 1975),
406 pp.
Chapter 6
BATH, M., Lateral Inhomogeneities of the Upper Mantle, Tectonophysics 2, 483-514
(1965).
BATH, M., and BENIOFF, H., The Aftershock Sequence of the Kamchatka Earthquake of
November 4, 1952, Bull. Seism. Soc. Amer. 48, 1-15 (1958).
BATH, M., and DUDA, S. J., Earthquake Volume, Fault Plane Area, Seismic Energy,
Strain, Deformation and Related Quantities, Ann. Geofis. 17, 353-368 (1964).
BELOUSSOV, Y. Y., Againsl the Hypothesis of Ocean-Floor Spreading, Tectonophysics 9
489-511 (1970).
BENIOFF, H., Earthquakes and Rock Creep, Bull. Seism. Soc. Amer. 41, 31-62 (1951).
BENIOFF, H., Circum-Pacific Tectonics, Pub!. Dom. Obs. Ottawa, 20(2), 395-402 (1957).
BENIOFF, H., Movements on Major Transcurrent Faults, Int. Geophys. Ser., Acad. Press
3, 103-134 (1962).
BENIOFF, H., Source Wave Forms of Three Earthquakes, Bull. Seism. Soc. Amer. 53,
893-903 (1963).
BENIOFF, H., Earthquake Source Mechanisms, Science 143, 1399-1406 (1964).
BEN-MEN AHEM, A., Radiation of Seismic Surface Waves from Finite Moving Sources,
Bull. Seism. Soc. Amer. 51, 401-435 (1961).
BONNIN, J., and DIETZ, R. S. (Editors), Present State of Plate Tectonics, Tectonophysics
38,168 pp. (1977) (contains papers presented at a symposium in Grenoble in 1975).
400 Literature Review
CAREY, S. W., The Expanding Earth - An Essay Review, Earth Sci. Rev. 11, 105-143
(1975).
Cox, A., Plate Tectonics and Geomagnetic Reversals (Freeman, San Francisco 1973),
702 pp.
EVISON, E E, On the Occurrence of Volume Change at the Earthquake Source, Bull. Seism.
Soc, Arner. 57, 9-25 (1967).
ISACKS, 8., OLIVER, J., and SYKES, L. R., Seismology and the New Global Tectonics,
J, Geophys. Res. 73, 5855-5899 (1968).
KHATTRI, K" Earthquake Focal Mechanism Studies - A Review, Earth Sci. Rev. 9,
19-63 (1973).
LENSEN, G. J., Principal Horizontal Stress Directions as an Aid to the Study of Crustal
Deformation, Publ. Dorn. Obs. Ottawa 24 (10 j, 389-397 (1960).
LE PICHON, x., Sea-Floor Spreading and Continental Drift, J. Geophys. Res. 73, 3661-
3697 (1968).
OLIVER, J., SYKES, L., and ISACKS, 8., Seismology and the New Global Tectonics, Tectono-
physics 7, 527-541 (1969).
OLSSON, R., Strain Energy Release, De/ormation Energy Release and Related Problems,
Riv. Ital. Geofis. 22, 341-352 (1973).
OLSSON, R., Earthquake Activity in the Kamchatka-Kurile Islands-Japan Region NOl'ember
6. 1958-November 30. 1970. Riv. Ital. Geofis. 23. 43-56 (1974).
OLSSON, R., Some Aftershock Sequences in the Japan-Kamch{.'tka Region, Scism. Inst.,
Uppsala, Rep. No. 3-76. 12 pp., 2 tables, 2 figs (1976).
ROUSE, G. E., and BISQUE, R. E., Global Tectonics and the Earth's Core, The Mines Mag.,
Mining Eng" 8 pp" March 1968.
SCHEIDEGGER, A. E., The Geometrical Representation of Fault-Plane Solutions of Earth-
quakes, Bull. Seism, Soc. Amer. 47, 89-110(1957).
SCHEIDEGGER, A. E., Recent Advances in Geodynamics, Earth-Sci. Rev. 1, 133-153
(1966).
STEFANSSON, R .• The Use 0/ Transverse Waves in Focal Mechanism Studies, Tectono-
physics 3,35-60 (1966).
STEFANSSON, R., Methods of Focal Mechanism Studies with Application to Two Atlantic
Earthquakes, Tectonophysics 3, 209-243 (1966).
UYEDA, S. (Ed.), Subduction Zones. Mid-Ocean Ridges, Oceanic Trenches and Geodyna-
mics. Tectonophysics 37,246 pp. (1977) (contains papers presented at a symposium
in Grenoble in 1975).
WILSON, J. T, (Ed.), Continents Adrift. Readings from Sci. Amer. 1952-72 (Freeman,
San Francisco 1972), 172 pp.
Chapter 7
ANDERSON, D. L., Recent Evidence COllceming the Structure and Composition of' the
Earth's Mantle, Phys. and Chern. Earth 6,1-131 (1966).
ANONYMOUS, Geodynamics Project: Det'eloplllent of a u.s. Program, Trans. Amer. Geo-
phys. Union 52, 396-405 (1971).
BATH, M., Seismic Exploration of the Earth's Cl'lIst - Recellf Del'elopmellfs, Geol. Foren.
Forhandl. 80,291-308 (1958).
BATH, M., Die COllrad-Diskontinuitiit, Freiberger Forsch.-Hefte C 101,5-34 (1961).
BATH, M., Crustal Structure in Icelalld amI Surrounding Ocean, ICSU Review 4, 127-133
(1962).
Literature Review 401
BATH, M., An Analysis 0/ the Time Term Method ill Re/raction Seismology. Seism. Inst.,
Uppsala, Rep. No. /0-76.22 pp., 2 figs (1976).
BATH, M., et a!., A Seismic Re/raction Investigation 0/ Superficial Granitic Layering.
Seism. Inst., Uppsala, Rep. No. 7-76. 21 pp., 3 tables, 7 figs (1976).
BROWN, R. J., Lateral Inhomogeneity in the Crust and Upper Mantle from P- Wave Ampli-
tudes. Pure and App!. Geophys. /01. 102-154 (1972).
BULLEN, K. E., The Earth's Demity (Chapman and Hall, London 1975),420 pp.
BULLEN, K. E., and HADDON, R. A. W., Evidence from Seismology and Related Sources
on the Earth's Present Internal Structure. Phys. Earth Planet. Interiors 2.342-349
(1970).
COOK, K. L., The Problem 0/ the Mantle-Crust Mix: Lateral Inhomogeneity in the Upper-
most Part 0/ the Earth's Mantle. Advances in Geophys. 9. 295-360 (1962).
FINDLAY, D. C, and SMITH, C H., Drilling for Scientific Purposes. Geo!. Surv. Canada,
Paper 66-13.264 pp. (1966).
GURWITSCH, I. I., Seismische Erkundung (Geest & Portig, Leipzig 1970), 699 pp.
GUTENBERG, B., PKKP, P'P', and the Earth's Core, Trans. Amer. Geophys. Union 32,
373-390 (1951).
HADDON, R. A. W., and BULLEN, K. E., An Earth Model Incorporating Free Earth
Oscillation Data, Phys. Earth Planet. Interiors 2, 35-49 (1969).
IBRAHIM, A. K., Leaking and Normal Modes as a Means to Determine Crust-Upper
Mantle Structure for Different Paths to Sweden, Bull. Seism. Soc. Amer. 59,
1695-1712 (1969).
IBRAHIM, A. K., Effects 0/ a Rigid Core on the Reflection and Transmission Coefficients
from a Multi-Layered Core-Mantle Boundary, Pure and App!. Geophys. 91,95 -1 13
(1971).
IBRAHIM, A. K., The Amplitude Ratio PcP/P and the Core-Mantle Boundary, Pure and
App!. Geophys., 91,114-133 (1971).
KANA!, K., On the Group Velocity 0/ Dispersive Sur/ace Waves, Bull. Earthq. Res. Inst_
Tokyo 29, 49-60 (1951).
KULHANEK, 0., and BROWN, R. J., P-Wave Velocity Anomalies in the Earth's Mantle
from the Uppsala Array Observations, Pure and Appl. Geophys. 112, 597-617
(1974).
LEE, W. H. K., and TAYLOR, P. T., Global Analysis 0/ Seismic Re/raction Measurements,
Geophys. J. 11.389-413 (1966).
LUBIMOYA, E. A., Thermal History o/the Earth with Consideration o/the Variable Thermal
Conductivity 0/ its Mantle. Geophys. J. I, 115-134 (1958).
MASON, 8., Principles o/Geochemistry (Wiley & Sons, New York 1952),276 pp.
MULLER, G., MULA, A. H., and GREGERSEN, S., Amplitudes 0/ Long-Period PcP and the
Core-Mantle Boundary. Phys. Earth Planet. Interiors 14. 30-40 (1977).
PRESS, F., Earth Models Consistent with Geophysical Data, Phys. Earth Planet. Interiors
3,3-22 (1970).
SANTO, T., Lateral Variation 0/ Rayleigh Wave Dispersion Character; I. Observational
Data, Pure and Appl. Geophys. 62,49-66 (1965).
SANTO, T., Lateral Variation 0/ Rayleigh Wave Dispersion Character; II. Eurasia, Pure
and Appl. Geophys. 62, 67-80 (1965).
SANTO, T., Lateral Variation 0/ Rayleigh Wave Di~persion Character; III. Atlantic Ocean,
A/rica and Indian Ocean, Pure and App!. Geophys. 63,40-59 (1966).
TOKSQZ, M. N., CHINNERY, M. A., and ANDERSON, D. L., Inhomogeneities in the Earth's
Mantle, Geophys. J. 13, 31-59 (1967).
Chapter 8
AL-SADI, fl. N., Dependence of the P-Wave Amplitude Spectrum on Focal Depth. Pure and
App!. Geophys. 104.439-452 (1973).
BATH, M., Futur developpement des rbeaux de stations seismologiques, Scientia 99, 184-
191 (1964).
BATH, M., Seismic Recording Possibilities in Sweden, FOA4, Stockholm, A 4466-4721,
34 pp. (1965).
BATH, M., Underground Measurements of Short-Period Seismic Noise, Ann. Geofis. 19,
107-117 (1966).
BATH, M., Methods to Improve Seismograph Data, Scientia 102, 304-317 (1967).
BATH, M., Spectral Analysis in Geophysics (Elsevier, Amsterdam 1974), 563 pp.
BEAUCHAMP, K. G. (Ed.), Exploitation of Seismograph Networks. NATO Advanced Study
Inst., E 11 (Noordhoff, Leiden 1975),647 pp.
BROWN, R. J., Slowness and Azimuth at the Uppsala Array. Part 1: Array Calibration and
Event Location. Pure and App!. Geophys. 105.759-769 (1973).
BROWN, R. J., Slowness and Azimuth at the Uppsala Array. Part 2: Structural Studies.
Pure and App!. Geophys. 109, 1623-1637 (1973).
CRAMPIN, S., and BATH, M., Higher Modes of Seismic Surface Waves: Mode Separation.
Geophys. J. 10. 81-92 (1965).
ENAYATOLLAH, M. A., Continental-Array Measurements of P- Wave Velocities in the Mantle.
Pure and App!. Geophys. 94. 136-147 (1972).
FARNBACH, J. S., The Complex Envelope in Seismic Signal Analysis. Bull. Seism. Soc.
Amer. 65,951-962 (1975).
FROSCH, R. A., and GREEN, P. E., JR., The Concept of a Large Aperture Seismic Array,
Proc. Roy. Soc. A 290, 368-384 (1966).
HUSEBYE, E. S., and JANSSON, B., Application of Array Data Processing Techniques to
the Swedish Seismograph Stations, Pure and App!. Geophys. 63, 82-104 (1966).
JANSSON, B., and HUSEBYE, E. S., Application of Array Data Processing Techniques
to a Network of Ordinary Seismograph Stations, Pure and App!. Geophys. 69,
80-99 (1968).
KORKMAN, K., Aftershock P-wave Spectra and Dynamic Features of the Aleutian Islands
Earthquake Sequence of February 4, 1965, Tectonophysics 5, 245-266 (1968).
KULHANEK, 0., P-wave Amplitude Spectra of Nevada Underground Nuclear Explosions,
Pure and App!. Geophys. 88, 121-136 (1971).
KULHANEK, 0., Signal and Noise Coherence Determination for the Uppsala Seismograph
Array Station. Pure and App!. Geophys. 109. 1653-1671 (1973).
KULHANEK, 0., Introduction to Digital Filtering in Geophysics (Elsevier, Amsterdam 1976),
168 pp.
LEONG, L. S., Crustal Structure of the Baltic Shield beneath Umed. Sweden. from the
Spectral Behavior of Long-Period P Waves. Bull. Seism. Soc. Amer. 65. 113-126
(1975).
SUTTON, G. H., McDONALD, W. G., PRENTISS, D. D., and THANOS, S. N., Ocean-Bottom
Seismic Observations, Proc. IEEE 53, 1909-1921 (1965).
Literature Review 403
WHITCOMB, J. H., Array Data Processing Techniques Applied to Long-Period Shear Waves
at Fennoscandian Seismograph Stations, Bull. Seism. Soc. Amer. 59,1863-1887
(1969).
WHITEW AY, F. E., The Use of Arrays for Earthquake Seismology, Proc. Roy. Soc. A 290,
328-342 (1966).
Chapter 9
ANDERSON, O. L., and LIEBERMANN, R. C., Sound Velocities in Rocks and Minerals,
VESIAC Rep. 7885-4-X, 182 pp. (1966).
ANONYMOUS, Symposium on Seismic Models, LibIice near Prague, November 9-12, 1965
(a collection of papers presented at the symposium), Stud. Geophys. Geodaet.
10, 239-400 (1966).
BRACE, W. F., Laboratory Studies Pertaining to Earthquakes, Trans. New York Acad.
Sci., Ser. II 31, 892-906 (1969).
DUDA, S. J., The Stress Around a Fault According to a Photoelastic Model Experiment,
Geophys. J. 9, 399-410 (1965).
LAVIN, P. M., and HOWELL, B. F., JR., Model Studies of Effects of Near-Source Velocity
Discontinuities on First-Motion Patterns, Bull. Seism. Soc. Amer. 53, 933-954
(1963).
PRESS, F., A Seismic Model Study of the Phase Velocity Method of Exploration, Geophysics
22, 275-285 (1957).
PRESS, F., Elastic Wave Radiation from Faults in Ultrasonic Models, Pub!. Dom. Obs.
Ottawa 20 (2),271-277 (1957).
PRESS, F., OLIVER, J., and EWING, M., Seismic Model Study of Refractions from a Layer
of Finite Thickness, Geophysics 19, 388-401 (1954).
RAMBERG, H., Gravity, Deformation and the Earth's Crust (Academic Press, New York
1967), 214 pp.
VANEK, J., Revised Amplitude Curves of Seismic Body Waves for the Region of South-
Eastern Europe, Stud. Geophys. Geodaet. 13, 173-179 (1969).
Chapter 10
ADAMS, R. D., The Haicheng, China, Earthquake of 4 February, 1975: The First Success-
fully Predicted Major Earthquake, Bull. New Zealand Nat. Soc. Earthq. Eng. 9,
32-42 (1976).
ANONYMOUS, ESSA Symposium on Earthquake Prediction, U.S. Dept. Commerce, 167 pp.
(1966).
ANONYMOUS, Proposal for a Ten-Year National Earthquake Hazards Program. Ad hoc
Interagency Working Group for Earthquake Research, Fed. Council Sci. & Tech.,
Washington, D.C., 81 pp. (1968).
BATH, M., Earthquake Prediction, Scientia 101,234-243 (1966).
BATH, M., Artificial Release of Earthquakes, Scientia 105, 19 pp (1970).
BATH, M., and WAHLSTROM, R., A Rockburst Sequence at the Griingesberg Iron are
Mines in Central Sweden, Seism. Inst., Uppsala, Rep. No. 6-76, 30 pp., 3 tables,
1 appendix, II figs (1976).
BLOT, c., Origine profonde des seismes superficiels et des eruptions volcaniques. IUGG
XII Gen. Ass. Seismology, 1963.
BOUCHER, G., RYALL, A., and JONES, A. E., Earthquakes Associated with Underground
Nuclear Explosions, J. Geophys. Res. 74, 3808-3820 (1969).
404 Literature Review
CALOI, P., and SPADEA, M. C., Principali risultati conseguiti durante l'osservazione geo-
dinamica, opportunamente estesa nel tempo, di grandi dighe di sbarramento, e loro
giustificazione teoriche, Ann. Geofis. 19, 261-286 (1966).
CARDER, D. S., Seismic Investigations in the Boulder Dam Area, 1940-1944, and the
Influence of Reservoir Loading on Local Earthquake Activity, Bull. Seism. Soc.
Amer. 35, 175-192 (1945).
COMNINAKIS, P., DRAKOPOULOS, J., MOUMOULlDlS, G., and PAPAZACHOS, B., Fore-
shock and Aftershock Sequences of the Cremasta Earthquake and their Relation
to the Waterloading of the Cremasta Artificial Lake, Ann. Geofis. 21, 39-71
(1968).
DUDA, S. J., Phanomenologische Untersuchung einer Nachbebenserie aus dem Gebiet der
Aleuten-Inseln, Freiberger Forsch.-Hefte C 132, 9-90 (1962).
DUDA, S. J., Strain Release in the Circum-Pacific Belt: Chile 1960, J. Geophys. Res. 68,
5531-5544 (1963).
DUDA, S. J., and BATH, M., Strain Release in the Circum-Pacific Belt: Kern County 1952,
Desert Hot Springs 1948, San Francisco 1957, Geophys. J. 7, 554-570 (1963).
FEDOTOV, S. A., et aI., Long- and Short-Term Earthquake Prediction in Kamchatka.
Tectonophysics 37.305-321 (1977).
FOWLER, R. A., Earthquake Prediction from Laser Surveying. NASA SP-5042. 32 pp.
(1968).
GUPTA, H. K., and RASTOGI, B. K., Dams and Earthquakes (Elsevier, Amsterdam 1976),
229 pp.
GUTENBERG, B., Seismology, Geol. Soc. Amer., 50th Ann. Vol., 439-470 (1941).
HAGIWARA, T., and RIKITAKE, T., Japanese Program on Earthquake Prediction, Science
157,761-768 (1967).
KNOPOFF, L., Earth Tides as a Triggering Mechanism for Earthquakes, Bull. Seism. Soc.
Amer. 54, 1865-1870 (1964).
MEAD, T. S., and CARDER, D. S., Seismic Investigations in the Boulder Dam Area in 1940,
Bull. Seism. Soc. Amer. 31, 321-340 (1941).
MJACHKIN, V. I., BRACE, W. F., SOBOLEV, G. A., and DIETERICH, J. H., Two Models for
Earthquake Forerunners. Pure and Appl. Geophys. 113. 169-181 (1975).
PAGE, R. (Ed.), Proceedings of the Second United States-Japan Conference on Research
Related to Earthquake Prediction. Nat. Sci. Foundation and Japan Soc. Prom.
Sci., 106 pp. (1966).
PRESS, F., Earthquake Prediction. Sci. Amer. 232. 14-23 (1975).
PRESS, F., and BRACE, W. F., Earthquake Prediction, Science 152, 1575-1584 (1966).
RIKITAKE, T., A Five-Year Plan for Earthquake Prediction Research in Japan, Tectuno-
physics 3, 1-15 (1966).
RIKIT AKE, T. (Ed.), Earthquake Prediction, Tectonophysics 6, 1-87 (1968).
RIKITAKE, T. (Ed.), Earthquake Mechanics, Tectonophysics 9, 97-300 (1970) (contains
papers presented at a symposium in Madrid in September 1969).
RIKITAKE, T. (Ed.), Focal Processes and the Prediction of Earthquakes. Tectonophysics
23.219-318 (1974) (contains papers presented at a symposium in Lima in 1973).
RIKITAKE, T., Earthquake Prediction (Elsevier, Amsterdam 1976),357 pp.
RYALL, A., and SAVAGE, W. U., A Comparison of Seismological Effects for the Nevada
Underground Test Boxcar with Natural Earthquakes in the Nevada Region. J.
Geophys. Res. 74. 4281-4289 (1969).
Literature Review 405
Chapter 11
GORDON, D. W. Travel Time Curve, ESSA Symp. Earthq. Pred., 141-145 (1966).
GRIGGS, D. T., ;nd PRESS F., Probing the Earth with Nuclear Explosions (Univ. of Calif.,
Lawrence Radiation Lab., 1959), 31 pp.
KOGAN, S. D., Travel Times of Body Waves from Surface Focus, Izv. Akad. Nauk SSR,
Ser. Geofiz. 3. 371-380(960). Also in IASPEI, Trav. Sci. A 21, 15-22 (1961).
KORKMAN (KOGEUS), K., A Synthesis of Short-Period P-wave Records from Distant
Explosion Sources, Bull. Seism. Soc. Amer. 58, 663-680 (1968).
KULHANEK, 0., Source Parameters of Some Presumed Semipalatinsk Underground Nuclear
Explosions. Pure and App!. Geophys. 102. 51-66 (1973).
MEYER, K., Secondary Pressure Pulses from Underwater Explosions. Seism. Inst., Uppsala,
Rep. No. 9-75, 17 pp., 5 tables, 7 figs (1975).
MEYER, K., and BATH, M., First-Arrival Seismic Waves Generated by Underwater Ex-
plosions. Seism. Inst., Uppsala, Rep. No. 2-75. 13 pp., 6 tables, 5 figs (1975).
MOROKHOY, I. D. (Ed.), Nuclear Explosions for Peaceful Purposes. Trans!. from the
Russian Atomnye vzryvy v mirnykh tselyakh, Atomizdat, Moscow, 124 pp.
(1970), U.S. Atomic Energy Commission, UCRL-Trans-J0517. 190 pp. (1971).
NORDYKE, M. D. (Ed.), French and Soviet Papers Presented at the Second Panel on the
Peaceful Uses of Nuclear Explosions. Vienna. Austria. 18-22 January 1971.
U.S. Atomic Energy Commission, UCRL-Trans-J0543. 116 pp. (1971). (There
are similar publications also on later conferences in Vienna on this topic.)
PERSEN, L. N., Rock Dynamics and Geophysical Exploration (Elsevier, Amsterdam 1975),
276 pp.
PRESS, F., DEWART, G., and GILMAN, R., A Study of Several Diagnostic Techniquesfor
Identifying Earthquakes, J. Geophys. Res. 68, 2909-2928 (1963).
ROBINSON, E. A., Statistical Communication and Detection with Special Reference to
Digital Data ProceSSing of Radar and Seismic Signals (Griffin, London 1967),
362 pp.
RODEAN, H. c., Understanding and Constructit'ely Using the Effects of Underground Nuclear
Explosions, Rev. Geophys. 6, 401-445 (1968).
RODEAN, H. C., Nuclear-Explosion Seismology (U.S. Atomic Energy Commission, 1971),
156 pp.
SPRINGER, D. L., and KINNAMAN, R. L., Seismic Source Summary for U.S. Underground
Nuclear Explosions. /961-1970. Bull. Seism. Soc. Amer. 61. 1073-1098 (1971).
SPRINGER. D. L., and KINNAMAN, R. L., Seismic Source Summary for U.S. Underground
Nuclear Explosions. 1971-1973. Bull. Seism. Soc. Amer. 65. 343-349 (1975).
Chapter 12
ALFYEN, H., Origin of the Moon, Science 148,476-477 (1965).
BASTIN, J. A., The Diffusion of Lunar Seismic Energy, Phys. Earth Planet. Interiors 4,
218-221 (1971).
BOWIN, C, SIMON, B., and WOLLENHAUPT, W. R., Mascons: A Two-Body Solution. J.
Geophys. Res. 80. 4947-4955 (1975).
BULLEN, K. E., Origin of the Moon. Nature 167. 29 (1951).
BULLEN, K. E., On the Internal Constitutions of the Planets. Proc. Astr. Soc. Australia 1.
1-6 (1967).
CHANG, G. K., GUNTHER, P., and JAMES, D. B., A Secondary Ejecta Explanation of a
Lunar Seismogram. J. Geophys. Res. 75. 7426-7438 (1970).
Literature Review 407
COOPER, M. R., KOVACH, R. L., and WATKINS, J. S., Lunar Near-Surface Structure.
Rev. Geophys. and Space Phys. 12.291-308 (1974).
DATTA, A. N .. On the Energy Required to Form the Moon. Monthl. Not. Roy. Astr. Soc.,
Geophys. Suppl. 6(9),535-539 (1954).
DERR, J. S.,Travel Times. Variational Parameters. and Love Numbers for Moon Models.
Bull. Seism. Soc. Amer. 60. 697-716 (1970).
DUENNEBIER. F., and SUTTON, G. H., Thermal Moonquake!>. J. Geophys. Res. 79.4351-
4363 (1974).
DUENNEBIER, F., and SUTTON, G. H., Meteoroid Impacts Recorded by the Short-Period
Component of Apollo 14 Lunar Passive Seismic Station. J. Geophys. Res. 79.
4365-4374 (1974).
EWING, M., et aI., Seismology of the Moon and Implications on Internal Structure. Origin
and Evolution. Highlights of Astronomy (De Jager, Ed.), 155-172 (1971).
KOVACH, R. L., and ANDERSON, D. L., The Interiors of the Terrestrial Planets, J. Geophys.
Res. 70,2873-2882 (1965).
KOZYREV, N. A., Volcanism on the Planets, Tectonophysics 1, 451-454 (1964).
LATHAM, G., EWING, M., PRESS, F., and SUTTON, G., The Apollo Passil'e Seismic Ex-
periment, Science 165, 241-250 (1969).
LATHAM, G. Y., et aI., Apollo 11 Passil'e Seismic Experiment, Proc. Apollo II Lunar
Science Conf. 3, 2309-2320 (1970). In abbreviated form also in Science 167,
455-457 (1970).
LATHAM, G. Y., McDONALD, W. G., and MOORE. H. J., Missile Impacts as Sources of'
Seismic Energy on the Moon, Science 168, 242-245 (1970).
LATHAM, G., et aI., Moonquakes. Science 174, 687-692 (1971).
MIDDLEHURST, B. M., An Analysis of Lunar Events, Rev. Geophys. 5, 173-189 (1967).
NAKAMURA, Y., and LATHAM, G. Y., Travel Times of Body Waves in the Moon, Bull.
Seism. Soc. Amer. 60, 63-78 (1970).
NASA (National Aeronautics and Space Administration), Apollo 15, Preliminary Science
Report. Spec. Publ. SP-289, 526 pp. (1972).
O'KEEFE, J. A., Origin of the Moon, J. Geophys. Res. 74, 2758-2767 (1969), and 75.
6565-6574 (1970).
SUTTON, G. H., and STEINBACHER, R., Surveyor Seismograph Experiment. J. Geophys.
Res. 72. 841-844 (1967).
TOKSDZ, M. N., et aI., Velocity Structure and Properties of the Lunar Crust, The Moon,
Reidel Publ. Co., 4, 490-504 (1972).
TOKS6z, M. N., DAINTY, A. M., SOLOMON, S. C, and ANDERSON, K. R., Velocity Structure
and Evolution of the Moon, Geochim. et Cosmochim. Acta, Suppl. 4, 3, 2529-
2547 (1973).
409
Subject Index
Continental crust 77, 227 Detection 269, 335, 337, 339, 394
Continental drift 185 Deterministic prediction 294
Continental path 229 Dextral displacement 173
Contraction, earth 172 Diffraction 65, 73, 278, 283
Convection current 172, 233, 236, 308 Diffusion 302
Coordinate 108 Digital seismograph 46
Core 23, 25, 70, 73, 87, 94, 185, 203, 215, Dilatancy 298
232, 368, 376 Dilatancy-diffusion 302
Correlation 257 Dilatation 62, 171, 257
Cosmic radiation 377 Dipole 175, 177
Coupling 348, 349 Dip-slip fault 173
Coupling factor 348, 349 Direct-recording seismograph 34, 35, 39
Crater 369 Discontinuity, Conrad 75, 222, 224
Crater experiment 348 Discontinuity, Mohorovicic see Mohoro-
Creep 196 vicic discontinuity
Cretaceous period 164 Dislocation 122, 278, 325
Critical angle of incidence 289 Dispersion 80, 81, 228, 230, 269, 371
Critical damping 32 Dispersion equation 89
Crust 94, 203, 220, 333, 376 Doppler principle 178
Crustal deformation 297 Double dipole 177
Crust, continental 77, 227 Drilling 221
Crust, moon 376 Dynamics of the earth 152
Crust, oceanic 77, 227
Curie point 235 Earth contraction 172
Curriculum, seismology 381 Earth current 301
Earth dynamics 152
Dam 128, 310,321, 392 Earth expansion 172
Dam construction 317 Earth magnetism 232, 235 see also
Damping 32 Geomagnetism
Damping, air 47 Earthquake acceleration 121, 127
Damping, aperiodic 32 Earthquake, artificial release 291
Damping, critical 32 Earthquake catastrophe 137
Damping, oil 47 Earthquake, deep 190, 303
Data exchange 245 Earthquake, destructive 139
Decoupling 348 Earthquake effect 138
Deep drilling 221 Earthquake engineering 28, 42, 127, 157,
Deep earthquake 190, 303 271, 393
Deep-sea trench 164 Earthquake forerunner 303
Deformation 31, 199 Earthquake frequency 153, 375
Deformation characteristic 199 Earthquake insurance 28, 295, 393
Deformation, crustal 297 Earthquake mechanism 169, 190, 269,
Deformation seismograph 31 284,310,327,341
Density 23, 203, 209, 232, 235, 236, 364 Earthquake, near 75, 221
Density gradient 210 Earthquake, number 150, 151, 152, 153,
Depth, focal 126, 167, 325, 338, 361 154,298
Destruction belt 151 Earthquake periodicity 202
Destructive earthquake 139 Earthquake prediction 267, 286, 291, 393
411
81, 84, 87, 88, 89, 94, 97, 101, 112, 178, Mining operation 28,310,323
231, 343, 346 Mode, fundamental 79, 87, 94
Low-velocity layer 78, 87, 91, 92, 206, Mode, higher see Higher mode
242, 278, 282, 283 Model, moon 375, 376
Lubrication 313 Model seismology 277
Lunar Orbiter 367 Model tectonics 278
Lunar tide 325, 374 Modified Mercalli scale (MM-scale) 123
Lunik III 366 Modulus, bulk 63
Modulus of compressibility 18
Macroseismic area 126 Modulus of elasticity 18
Macroseismic observation 22, 125 Modulus of incompressibility 63
Macroseismic questionnaire 133 Modulus of rigidity 18, 63, 122, 213
Magnetic reluctance 44 Modulus, shear 63
Magnetism, earth 232, 235 Mohole project 221
Magnetism, moon 365 Mohorovicic discontinuity (Moho) 19, 25,
Magnetostriction 54 75, 203, 220, 221, 224, 240, 262, 283,
Magnification 34 331,360,363
Magnification curve 48 Moment of inertia 209, 211, 214
Magnitude 105, 114, 121, 197, 267, 268, Moment, seismic 122, 325
269, 318, 346, 350, 351 Moon 299, 364
Magnitude formula 120 Moon crust 376
Mantle 94, 203 Moon magnetism 365
Mantle, moon 376 Moon mantle 376
Mantle wave 84,94, 181 Moon model 375, 376
Maria 346 Moonquake 369, 370, 374
Mars 239, 364, 370, 378, 379 Moon research 363
Marsquake 379 Moon-scraper 366
Mascon 377 Moon seismograph 370
Maxwell resistance bridge 51 Moon slide 371
Mechanical-optical seismograph 39 Moon temperature 369, 377
Mechanical seismograph 38 Moon topography 365
Mechanism, earthquake 169, 191, 269, Moscow-Prague formula 119
284, 310, 327, 341 Mountain formation 180
Mechanism, explosion 341 Mountain root 228
Megaton 349 Moving-coil seismograph 40
Mercalli scale 123, 124 MSK intensity scale 123, 124, 130
Mercury 239 Multiple station see Array station
Mesosphere 187
Metallic phase 237 Nanometre 372
Meteor 369, 371 National Aeronautics and Space Admi-
Meteorite 232, 236, 239 nistration (NASA) 407
Microearthquake 42, 156, 248, 274, 299 National Earthquake Information Service
Microseisms 99,249, 259, 266, 371 (NElS) see U. S. Geological Survey
Mid-Atlantic belt (ridge, rift) 160, 161, 184 National Science Foundation 292
Migration 304 Near earthquake 75, 221
Milne-Shaw seismograph 35, 39 Neptune 235, 239
Mine 252, 323 Nevada Test Site (NTS) 333
414
Seismogram 29 Shear 18
Seismograph 24, 29 Shear modulus 63
Seismograph, azimuthal 31 Shear stress 62, 199, 301
Seismograph, Benioff 35, 43 Shear wave see S-wave
Seismograph, bore-hole 55, 253 Shelf 228
Seismograph, Columbia 43 Shield 162
Seismograph, deformation 31 Shock wave 277
Seismograph, digital 46 Signal sensitivity 251
Seismograph, direct-recording 34, 35, 39 Signal-to-noise ratio 255, 257
Seismograph, electromagnetic 34 Silent earthquake 298
Seismograph, electrostatic 45 Single dipole 175
Seismograph, Galitzin 35, 40 Sinistral displacement 173
Seismograph, Grenet-Coulomb 41,249 SKIKP 74
Seismograph, long-period 229 SK1KS 74
Seismograph, mechanical 38 SKJKP 74
Seismograph, mechanical-optical 39 SKJKS 74
Seismograph, Milne-Shaw 35, 39 SKP 71
Seismograph, moon 370 SKS 71, 97, 246
Seismograph, moving-coil 40 Slow motion 128, 297
Seismograph, ocean-bottom 55, 77, 265 Sn 77, 93, 94, 224
Seismograph, pendulum 38 Snell's law 64
Seismograph, portable 42 SOFAR channel 336
Seismograph, Press-Ewing 42 Solar wind 378
Seismograph, reluctance 43 Solid-earth physics 14
Seismograph, rotation 31 Solid mechanics 277
Seismograph, strain 52, 179 Solid state physics 277
Seismograph, strong-motion 42 Sonic boom 125
Seismograph, torsion 39, 115 Source parameter 105, 223
Seismograph, translation 31 sP 69
Seismograph, triaxial 31 SP 68
Seismograph, Wiechert 34, 38 Spectral analysis 260
Seismograph, Wood-Anderson 39, 115, Spectrum 122, 127, 180, 273, 338, 342
Il7 Spheroidal vibration 85, 87, 94
Seismological Institute, Uppsala 56, 90, sPP 69
119, 154, 201, 242, 245, 246, 247, 248, SPP 68
249, 381, 395 Spreading, geometrical 114
Seismometer 29 Spreading, ocean-floor 186
Seismos 13 sPS 69
Seismoscope 16,29 SPS 68
Selenodesy 377 sS 69, 176, 375
Selenophysical phenomena 369 SS 68, 176, 375
Selenothermy 377 sSP 69
Sensitivity, signal 251 SSP 68
Sg 77, 78, 93, 94, 224, 346, 359, 360, 361 sSS 69
SH 62, 64, 66, 88, II 2, 175,258 SSS 68
Shadow zone 70,91 Station correction, magnitude Il4
Shaking table 49 Station correction, travel time 261
417
VELA Project 27, 329 Wave, guided see Guided wave and Chan-
Velocity 63 nel waVe
Velocity dispersion 80,228,230 Wave, longitudinal see P-wave
Velocity filtering 259 Wave, Love see Love wave
Velocity gradient 215 Wave, mantle 84, 94, 181
Velocity, group 81, 228,229,230 Wave parameter 65
Velocity, phase 81, 228, 281 Wave, Rayleigh see Rayleigh wave
Velocity, P, S 206, 208, 376 Wave, shear see S-wave
Velocity, wave 63, 81, 286, 302 Wave, shock 277
Venus 239, 364, 370, 378, 379 Wave, Stoneley 61, 80
Vibrating string 88 Wave, surface see Surface wave
Vibration, free see Free vibration Wave, tidal 21, 367
Vibration measurement 130,392 Wave, transformed 69, 288
Viking 379 Wave, transverse see S-wave
Volcanic earthquake 21, 163, 194, 375 Wave velocity 63, 81, 286, 302
Volcanic eruption 163,297,299,369 Wiechert iron-core hypothesis 232
Volcanism 164, 190,369 Wiechert seismograph 34, 38
Volcano 23
Volcanology 14 Wiederkehr-Welle 85
Volume, earthquake 197 Wood-Anderson seismograph 39, 115, 117
Volume, planetary 364 World map, seismic 2,22, 159
World network 262,270
Water injection 315 Wrench fault 173
Wave, body see Body wave
Wave, channel see Channel wave Yield 348
Wave form 343
Wave guide 278 see also Low-velocity layer Zoning map, seismic 157, 293
419
Author index
Doppler, Chr. 178 207, 211, 218, 224, 235, 246, 292, 384,
Dorman, J. 2 388, 395, 397, 398, 401, 404
Drakopoulos, J. 404
Dreimanis, E. 6 Haa1ck, H. 240
Duda, S. J. 154, 166, 184, 197, 199, 201, Haddon, R. A. W. 214,215, 397, 401
277, 305, 399, 403, 404 Hagiwara, T. 291, 404, 405
Duennebier, F. 407 Hast, N. 54, 184,299,317,392,396
Heezen, B. C. 161,399
Elsasser, W. M. 243 Heiland, C. A. 384
EnayatoIlah, M. A. 397, 402 Herrin, E. 99, 397
Eucken, A. 235 Hiller, W. 41
Evernden, J. F. 405 Hoernes, R. 19, 20
Evison, F. F. 400 Holmes, A. 383
Ewing, J. A. 19,24 Hooke, R. 17, 19, 62, 196
Ewing, W. M. 42, 43, 49, 56, 57, 73, 83, Housner, G. W. 398
86, 90, 161, 283, 384, 399, 403, 407 Howell, B. F., Jr. 285, 384, 403
Husebye, E. S. 50, 257, 397, 398, 402
Farnbach, J. S. 402,405
Fedotov, S. A. 404 Ibrahim, A. K. 217,231,242,401
Findley, D. C. 401 lsacks, B. 186, 191, 400
Forel, F. A. 22, 123, 124
Foucault, L. 47 James, D. B. 406
Fourier, J. B. J. 38 Jansson, B. 50, 257, 402
Fowler, R. A. 404 Jardetzky, W. S. 384
Freeman, J. R. 398 Jeffreys, B. S. 395
Frosch, R. A. 402 Jeffreys, H. 95, 96, 97, 98, 99, 204, 205,
206, 207, 208, 210, 239, 246, 330, 331,
Galileo, G. 17, 19 384, 387, 388, 395, 397
Galitzin, B. 19, 24, 26, 35, 40, 41, 42, 45, Johnson, D. P. 41
47,48,49, 51, 57, 102 Jones, A. E. 403
Gane, P. G. 45 Jones, D. A. III
Garmany, J. G. 405
Gasparini, P. 395 Kanai, K. 228, 401
Gilman, R. 346, 406 Klirnik, V. 123, 398
Gordon, D. W. 406 Kelvin, Lord 23
Gray, T. 19, 24 Kharin, D. A. 41
Green, P. E., Jr. 402 Khattri, K. 400
Gregersen, S. 401 King, D. W. 397
Grcnet, G. 41,47,48,50,57, 58,249,360, Kinnaman, R. L. 355, 406
396 Kirnos, D. P. 41
Griggs, D. T. 406 Knopoff, L. 404
Gunther, P. 406 Knott, C. G. 18, 19
Gupta, H. K. 404 Kogan, S. D. 330, 406
Gurwitsch, I. I. 401 Korkman (Kogeus), K. 402, 406
Gutenberg, B. 5, 19,23,25,41,42,70,76, Kovach, R. L. 407
84, 87, 91, 92, 95, 98, 99, ll5, ll6, ll7, Kozyrev, N. A. 407
119, 157, 159, 164, 191, 192, 205, 206, Kuhn, W. 233, 234, 235, 236, 237, 243
421
Geographical index
Yakutat Bay 140, 146, 147, 179, 181 Yunnan 141, 149
Yokohama 20, 138, 141
Yugoslavia 137, 147, 160, 291 Zagros 148
Yukon 145 Zambia 314,392
Yungay 148 Zurich 118, 119, 355