See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/247033215

On the Correlation of Seismoscope Response with Earthquake

Magnitude and Modified Mercalli Intensity

Article in Bulletin of the Seismological Society of America · April 1975

DOI: 10.1785/BSSA0650020307

CITATIONS READS

245 136

2 authors, including:

Mihailo Trifunac
University of Southern California
287 PUBLICATIONS 10,127 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mihailo Trifunac on 21 August 2014.

The user has requested enhancement of the downloaded file.

 Bulletin of the
Seismological Society of America
Vol. 65 April 1975 No. 2

ON CORRELATION OF SEISMOSCOPE RESPONSE WITH EARTHQUAKE

MAGNITUDE AND MODIFIED MERCALLI INTENSITY

BY M. D. TRIFUNACAND A. G. BRADY

ABSTRACT
A quantitative measure of the Modified Mercalli Intensity Scale for earth-
quakes in ~he western United States has been developed by correlating the peak
seismoscope relative displacement response, Sd, with the reported site intensity,
IMM. This correlation can be approximated by

1 100.288IMM

for IMM =< VIII and is characterized by one standard deviation of about 0.7 Sa.
The data used in this study do not indicate an obvious type of dependence of Sa on
local site conditions.
A method for computing the analog of the local earthquake magnitude,
Mseismoscope, has been presented for possible use in strong-motion seismol-
ogy and for sealing earthquakes by close-in measurements, when other seismo-
logical instruments may go off scale.

INTRODUCTION
Early notable attempts to develop simple strong-motion recording instruments that
would provide the structural engineer with information on response spectrum amplitudes
have been carried out by Galitzin (1913), Kirkpatrick (1927), Suyehiro (1926), and the
U.S. Coast and Geodetic Survey (Ulrich, 1941). Based on these pioneering ideas and
motivated by the increasing need for simple and inexpensive strong-motion recorders,
the U.S. Coast and Geodetic Survey and the California Institute of Technology devel-
oped the modern version of the strong-motion seismoscope (Hudson, 1958). Similar
instruments were concurrently developed and deployed in the U.S.S.R. (Medvedev,
1965) and later in India (Krishna and Chandrasekaran, 1965) and several other countries.
In 1972, over 1,200 seismoscopes were reported to have been installed in at least 16
countries (Fournier d'Albe, 1973), with the largest concentrations being in the U.S.A.
(400), Yugoslavia (320), and the U.S.S.R. (197).
During the last 15 years, numerous seismoscopes registered strong ground motion.
These measurements were used to infer the overall response spectrum amplitudes (e.g.,
Cloud and Hudson, 1961), to study the variability of strong ground motion with distance
from an earthquake source and site conditions (e.g., Hudson and Cloud, 1967; Hudson,
307
308 M. D. TRIFUNAC AND A. G. BRADY

1971), and to fill in the detailed information on strong ground motion where accelero-
graphs malfunctioned (Trifunac and Hudson, 1970) or were not available (Scott, 1973).
In the U.S.S.R., seismoscope response has also been used for correlations with and
determinations of intensity of ground shaking (Medvedev, 1965).
The purpose of this paper is to explore and to suggest the possibility of an extended
usage of this simple and rugged instrument. The motivation for our effort results from the
fact that the number of installed seismoscopes is already sufficiently large and can pro-
vide significant input to the overall problem of scaling and interpreting strong ground
motion.
We begin by developing correlations of seismoscope response with earthquake magni-
tude and local site conditions. We show that on the basis of such correlations seismo-
scope records might be used to infer the local earthquake magnitude in the epicentral
region where other more sensitive seismological instruments normally go off scale.
Subsequently, as proposed by Medvedev (1965), we develop correlations between the
local estimates of earthquake intensity and seismoscope peak response, and in this way
derive the approximate scaling laws for response spectra and Modified Mercalli intensities.
To avoid possible constraints on the above-mentioned correlations that might result
from~the limited amplitude range that can be recorded on the standard seismoscope
recording glass ( S a ~ 7.5 cm) and to increase the number of usable correlation points,
we derive all correlations from the computed values of 186 seismoscope responses
obtained from recorded strong-motion accelerograms.

COMPUTATION OF SEISMOSCOPE RESPONSE

The elementary theory of the response and the physical characteristics of the seismo-
scope (Figure l) have been presented in previous papers (Hudson, 1958; Cloud and
Hudson, 1961; Hudson and Cloud, 1967; Trifunac and Hudson, 1970) and will not be
repeated here. Laboratory and field tests have further shown that seismoscope response,
for small amplitudes, can be approximated by a system of two uncoupled second-order
differential equations, that the effect of vertical accelerations can be neglected, and that
the damping can be modeled by a constant equivalent viscous dash-pot (Trifunac and
Hudson, 1970).
For large deflections of the seismoscope pendulum, however, more refined theory is
required, and the Coulomb-type friction between the smoked glass and the recording
needle and the effects of vertical accelerations may be considered. If one neglects the
contributions of torsional and higher modes to the calculated response, then the non-
linear, coupled differential equations for the deflection angles ¢pand ~ (Figure 2) become
~+2on(~b+con 2 sin ~0cos ~ = -(ogn2/g) (cos ~0~:+sin cp;~) (la)
~ / + 2 0 9 n ( ~ +¢.On2 sin ~ cos cp -- --(O)n2/g) (COS~by+sin @~). (lb)
In these equations con is the natural frequency of the seismoscope pendulum normally
equal to 8.38 rad/sec, ( is the fraction of critical damping for equivalent viscously
damped systems, and g is the acceleration of gravity. 2, y and ;~ represent the absolute
components of acceleration of the pendulum support. For accurate response calculations
the ( dependence on the recorded amplitude is as shown in Figure 3. This functional form
has been confirmed by many laboratory experiments (Hudson, 1958). It represents a
superposition of the viscous amplitude-independent (5 to 7 per cent of critical) damping
resulting from the pendulum mass moving in the magnetic field (Figure 1) and the
equivalent viscous damping coefficient, which results from Coulomb friction between the
glass and the needle, and is inversely proportional to the response amplitude.
 CORRELATION OF SEISMOSCOPE RESPONSE WITH M AND IMM 309

FIG. 1. Photograph of the Wilrnot seismoscope with protective cover removed.

Z (t)

Y Y(9

FtG. 2. Coordinates t# and ~ used in describing the relative seismoscope motion. The general motion
of the seismoscope support is described by the general displacement x(t), y(t), z(t).
z 141 ~ I i I I t t , I T t I I I I I t-
E_
<~
a<~j 1 2

~ tC
cT -
O

la_
O _ ~

t.U
O
Fr
k.~
a. J I I I I r t I I [ I t i t I i iI
0 o.5 1.0 1.5
AMPLITUDE, CENTIMETERS (ON GLASS)
FIG. 3. Functional dependence of the fraction of critical damping on the amplitude of the seismoscope
response measured on the recording glass.
310 M. D. TRIFUNAC AND A. G. BRADY

The distance between the midpoint of the pivot spring suspension and the top surface
of the spherical standard watch glass is about 6.00 cm. This distance is required in con-
verting the angular deflections of the pendulum into the recorded amplitude on the
spherical glass surface.
From the analogy of the differential equations describing the relative motion of a
single degree-of-freedom viscously damped oscillator which is used in computations of
relative displacement response spectra, Sd, and the equations for small amplitude
seismoscope response, it can be shown that (Hudson and Cloud, 1967)
gT z
Sd = ~ 2 ~. . . . (2)

where Tis the natural period of the seismoscope pendulum, normally 0.75 sec, and where
:~. . . . measured in radians, is the largest angular deflection of the seismoscope pendulum
relative to the z axis(Figure 2) during a particular excitation. The value of :t.... can be
computed from the time histories (0(t) and 6(t). For large ¢p and ~ when nonlinear terms
in equations (la) and (lb) become significant, equation (2), of course, ceases to corre-
spond to its linear displacement spectrum analog.
Although valid for large pendulum deflections ~0 and 6, equations (la) and (lb) neglect
the contributions to the seismoscope response that may result from torsional, second
(Scott, 1973), and higher modes of vibration of the seismoscope pendulum. However, in
computing the synthetic seismoscope responses in this paper, we shall neglect those
torsional and higher mode contributions because the comparison between the calculated
and actually recorded seismoscope responses appears to be quite good. This has been
demonstrated on several occasions (e.g., Figures 9 a n d 10 of Hudson and Cloud, 1967)
when both a seismoscope and an accelerograph recorded strong ground motion at the
same point permitting direct comparison of the recorded and computed seismoscope
responses.

CORRELATION OF SEISMOSCOPE RESPONSE AND MODIFIED MERCALLI INTENSITY

Earthquake intensity scales are of necessity only descriptive and qualitative measures
of vibrational effects caused by earthquakes. These scales depend on numerous factors
that can change from one area to another. For example, the material and the type of
construction used for residential, industrial and public buildings, the type of geological
environment, the time of day when the earthquake occurs, public awareness of an existing
earthquake hazard, the methods used in collecting and interpreting intensity data (e.g.,
Lomnitz, 1970), all can result in significant differences of intensity determination in
various parts of the world. One possible remedy for these variations would be the estab-
lishment of some instrumental basis for intensity determinations.
One instrumental basis for the MKS intensity scale, now used in the U.S.S.R., has
been proposed by Medvedev (1965). To this end a conical pendulum, with static magni-
fication of 1.1, has been developed. This instrument, called an SBM seismometer, is very
similar in construction to the Wilmot type seismoscope used in the U.S. (Hudson, 1958).
It records on a spherical smoked glass, has a natural period of T = 0.25 sec, and the
fraction of critical damping ( = 0.08. By recording ground motions caused by earth-
quakes, explosions and machine vibrations, the empirical correlation between the MKS
intensity scale (Medvedev and Sponheuer, 1969) and the maximum recorded amplitude
on the SBM seismometer has been established (Medvedev, 1965).
To develop similar correlations between the Modified Mercalli intensity and the
maximum amplitude on the Wilmot Seismoscope (T = 0.75 sec, ( ~ 0.10), we calculated
 C O R R E L A T I O N OF SEISMOSCOPE RESPONSE W I T H M A N D IMM 311

synthetic seismoscope responses for 186 strong-motion accelerograph records (Trifunac

and Brady, 1975) now available in digital form. These accelerograms represent the best
strong-motion data recorded between 1933 and 1971 in the western United States and
are representative of earthquakes with magnitudes ranging from M = 4 to about M =
7.5. The Modified Mercalli intensity levels at the recording sites were estimated from the
isoseismal maps produced in the annual publication "United States Earthquakes" by
the U.S. Department of Commerce. Where detailed intensity levels were indicated in
these maps close to a recording site, these levels took precedence over the isoseismals. The
resulting intensity levels range from Ill to X, although the 186 accelerograph records
provide adequate data only for the intensities V, VI and VII.

TABLE 1

AVERAGE SEISMOSCOPE RESPONSE COMPUTED FROM 186

ACCELERATION RECORDS

o
Modified One Standard
Mercalli ,~a Deviation No.
Intensity (cm) of Sd (cm) of Data

Ili 0.30 -- 1*
IV 0.43 0.19 3
V 0.52 0.40 34*
VI 1.14 1.03 66*
VII 2.55 1.52 75
VIII 3.63 2.06 6
IX -- -- --
X 23.5 -- 1

Total 186

*In Table 3 of Trifunac and Brady (1975) the number

of data listed for intensities III, V and VI are 2, 33 and 67,
respectively,rather than those listed above. These changes
reflect the latest improvements in the classification of the
basic data-set and in no way affect the results and infer-
ences presented in this or our previous paper.

Most of the above 186 accelerograph records have been registered on alluvium or in a
"soft" geological environment (117 of 186 or 63 per cent) or on "intermediate" sedi-
mentary type rocks (54 of 186 or 29 per cent). Only a few records are available from
strong-motion stations located on "hard" rock (15 of 186 or 8 per cent). For simplicity
of notation in this paper, the data or the derived results that correspond to the stations
recording on "soft," "intermediate," or " h a r d " rocks will be labeled by the arbitrarily
chosen symbols 0, 1, and 2, respectively. The simplified geological data used to select
these 0, 1, and 2 labels and the methods employed in their selection have been presented
in our previous paper (Trifunac and Brady, 1975).
Table 1 and Figure 4 give the average maxima of the seismoscope responses and their
standard deviations for different Modified Mercalli intensities. It is seen that both ga and
its standard deviation rapidly increase for higher intensities. As already mentioned, the
number of data points is adequate to define Sa for the Modified Mercalli intensities V,
VI and VII only. However, to present all available information, we included even the
single point measurements in Figure 4.
It should be noted here that Table 1 and Figure 4 present the S a data versus the Modi-
fied Mercalli intensity, which is normally reported in terms of integer values, only for
312 M. D. TRIFUNACAND A. G. BRADY

convenience in computations. Since the seismoscope response Sa is an instrumental

measurement, it represents a more accurate quantity than a Modified Mercalli intensity
level and the scatter of S a versus the Modified Mercalli intensity in Table 1 and Figure 4
most probably is due to the poor accuracy of the intensity scale levels.
The functional relationship between the average Sn seismoscope reading and the
Modified Mercalli intensity cannot be derived from basic physical principles because the
earthquake intensity scale has no physical units. It has been derived by adopting a set of
conventions and definitions which give a qualitative but not quantitative description of
the level of shaking. By developing the empirical correlations of the type proposed by
Medvedev (1965), it may eventually become possible to "calibrate" various intensity
] I 1 I I I I I I I I
• a2
N3
f'd
o COMPUTED FROM 186
ACCELEROGRAPH RECORDS
• OBSERVED SEISMOSCOPE DATA
FROM SAN FERNANDO EARTH-
QUAKE, FEB. 9, 1971
1 ,~0.288
e ga (era) ~ 4~7~.2tu
INtM.

E
o
I
J

I I ~ I 1 l I I I
I ]I TIT ~T 37: :2I: STIT "GRIT 77" :X: "XT :X]I
I M M - M O D I F I E D MERCALLI I N T E N S I T Y

FIG. 4. Average maxima of both the computed and observed seismoscoperesponses and their standard
deviations for different Modified Mercalli intensities (see Tables I and 2).

scales by assigning to each intensity level a certain range of possible amplitudes recorded
on seismoscopes, accelerographs, seismometers or other related instruments. The data
we have at our disposal to date are certainly not adequate for this purpose but can be
used to derive the preliminary and approximate correlations. One such correlation that
can be used for preliminary prediction of Sd could be of the following form
1
Sa(cm) ~ ~ 10°'2881MM; |MM <VIII, (3)

where IMM represents the Modified Mercalli intensity. The data of Table 1 suggest that
the average standard deviation of S a would be of the order of 0.7 Sa.
To test the validity of the simplified seismoscope model (Figure 2) and the methods of
integrating the equations (la) and (lb) and thus the quality of the derived correlations in
Figure 4, Table 1 and our subsequent analysis, we correlated 116 seismoscope readings
 CORRELATION OF SEISMOSCOPE RESPONSE WITH M AND IMM 3 !3

(Hudson, 1971), which were recorded in the greater Los Angeles area during the San
Fernando, California, earthquake of February 9, 1971, with the corresponding Modified
Mercalli intensity. The results of this correlation are given in Table 2 and Figure 4. As
may be seen from Figure 4, the agreement between the trends indicated by the measured
and calculated seismoscope responses is very good. We interpret this to mean that our
method of computing the seismoscope response can be used as a substitute for actual
seismoscope recordings. This is clearly a useful result, since it allows us to derive numer-
ous correlations involving seismoscopes in the absence of actual seismoscope recordings.
In developing the above correlations (Tables 1, 2, Figure 4), we tabulated the Sa
readings irrespective of the local site and geological conditions at the recording stations.
Since the peaks of strong ground motion appear to be influenced by the type of site
conditions (Trifunac and Brady, 1975), it is appropriate to examine to what extent those
conditions may also be reflected in the seismoscope records computed from the same list
of 186 accelerograph records. The results of such analyses have been presented in Table 3
and Figure 5, where S e and its standard deviation have been tabulated versus Modified
Mercalli intensity and for different site classifications. Perusal of Table 3 shows that the
number of data used in this analysis is barely adequate to suggest the possible trends of
Sa and its standard deviation for intensities V, VI and VII and shows that many more
recordings will be required before we can develop reliable inferences on what effects the
site conditions may have on the recorded seismoscope response. Thus, the data plotted
in Figure 5 may eventually merge into a well-defined trend that might result from the
effects of geological conditions at the recording site, but does not, at this time, indicate
any obvious trends. In fact, some variations of Sd and its standard deviation in Figure 5
and Table 3 may be accidental and result from averaging over a small number of points
which are not representative of the whole population.

CORRELATIONS OF SEISMOSCOPE RESPONSE WITH MAGNITUDE AND

SITE CONDITIONS

Typically, an earthquake magnitude is proportional to the logarithm of the peak

amplitude of a seismogram recorded by a standard instrument. Starting from this
elementary definition, numerous magnitude scales have been developed for different
source-to-station distances, different types of recorded waves and for a variety of instru-
ments employed (e.g., Richter, 1958).
At close distances, typically less than 50 km, the amplitudes of ground motion caused
by even moderate earthquakes drive the conventional seismic instruments off scale and
the computation of magnitude must depend on more distant recordings. This leads to
several problems, the most important one being that in some cases no quick and local
information on the size of an earthquake may he available when the rescuing authorities
need it most. Therefore, it is especially fortunate when a strong-motion aceelerograph is
located so as to record near-field strong ground motions and thus provide statistics on the
local magnitude characteristics.
One simple and inexpensive solution to this problem seems to be the use of the seismo-
scope. Although only sensitive to a narrow band of input frequencies centered about its
natural frequency of 1.33 Hz, the seismoseope offers additional advantages in that it
represents a simple building model and thus also provides direct information on the
amplitude of response spectrum curves. With this in mind, we develop the correlations
between the peak seismoscope relative displacement response Sa and the magnitude
scale by using 182 accelerograph recordings obtained from 57 earthquakes in the Western
United States (see Table 2 of Trifunac and Brady, 1975).
314 M.D. TRIFUNAC AND A. G. BRADY

TABLE 2

AVERAGE OBSERVED SEISMOSCOPE DATA FROM S.F.

EARTHQUAKE

Gsd
Modified One Standard
Mercaili Sa of Deviation No.
Intensity (crn) Sa (cm) of Data

VIII 6.70 -- 1
VII 2.34 1.07 40
VI 1.20 0.84 52
V 0.34 0.28 23

Total 116

TABLE 3

AVERAGE SEISMOSCOPE RESPONSE COMPUTED FROM 186

ACCELERATION RECORDS AND FOR DIFFERENT SITE
CLASSIFICATIONS*

tr
Modified One Standard
Mercalli ~ Deviation No.
Intensity (cm), of Sa (cm) of Data

III-0 0.30
III-1
III-2
IV-0 0.70
IV-I 0.3O
IV-2
V-0 0.55 0.45 17
V-1 0.47 0.31 15t
V-2 0.60 0.50 2
VI-0 1,02 0.80 43
VI-I 1.35 1.30 16~
VI-2 1.41 1.35 7
VII-0 2.40 0.98 49t
VII-1 3.01 2.31 21t
VII-2 2.02 1.14 5
VIII-0 3.63 2.06 6
VIII-1
VIII-2
X-0
X-I
X-2 23.5

Total 186

*See Trifunac and Brady (1975).

t i n Table 5 of Trifunac and Brady (1975) the "number
of data" listed for intensities V-I, VI-I, VII-0 and VII-1
are 14, 17, 50 and 20 rather than those shown above.
These changes reflect the latest improvements in the
classification of the basic data-set and in no way affect the
results and inferences presented in this or our previous
paper.
 CORRELATION OF SEISMOSCOPE RESPONSE WITH M AND IMM 315

For most earthquakes studied in this paper, the magnitude determination has been
arrived at by using the standard definition of the local Richter magnitude M L (Richter,
1958)

M L = loglo A - l o g l o Ao(A), (4)

where A is the amplitude in millimeters recorded on the standard Wood-Anderson seismo-
graph whose natural period is 0.8 sec, static magnification Vs = 2800, and the fraction of
critical damping is 0.8 to 1.0. The empirically determined A o(A) represents the amplitude
in millimeters with which a standard seismometer would register an earthquake of
magnitude zero. Table 4, reproduced from Richter (1958), gives log~o Ao(A ) versus

7-
)II zlo,:] 1o , zi
SITE CLASSIFICATION
6- O-SOFT
I-INTERMEDIATE
2-HARD
5-

E4-
o
I
I i
5-
r
2-

z° ~' t
o i o
l Iv ]Z ::2] -I ~ ' r ~_-rrrr ' T
x-,

TM.M.- MODIFIED MERCALLI INTENSITY

FIG. 5. Average m a x i m a of seismoscope response c o m p u t e d from 186 acceleration records a n d their
s t a n d a r d deviations, for different Modified Mercalli intensities a n d for different site classifications
(see Table 3).

distance. In analogy with the definition (4), we can propose the magnitude derived from
seismoscope record, Mseismoscope,to be

Msei. . . . . . p~= log 1o Sd(cm) + 1 - log 1o A o(A) + log 1o 2800 -

- l o g 1 o (seismoscope transfer function). (5)
it is convenient to incorporate the effects of scaling and the transfer function into one
seismoscope scaling factor defined by
log 1o Sao = log1 o (seismoscope transfer function) - 1 - l o g 1 o 2800.
For a lightly damped seismoscope with the fraction of critical damping equal to about
0.10 and for a steady state sinusoidal excitation, the amplitude of the seismoscope trans-
fer function at its natural frequency would be 1/2~ ~ 5. This would then give

log ~o Sao = log I o Sd(Cm) -- Msoi. . . . cope -- log 1o A o(A) ~ - 3.75. (6)
To calculate the lOglo Sao for different magnitudes and recording site classifications,
we substituted the reported earthquake magnitudes for Ms~ismo~cope in equation (6),
computed log1 o Sd(Cm) from the maximum of the synthetic seismoscope response, and
316 M. D. TRIFUNAC AND A. G. BRADY

c o m p u t e d log~o Ao(A) f r o m Table 4 by using the k n o w n epicenter to station distance.

F o r the four m a g n i t u d e intervals M = 4.0 to 5.0, 5.0 to 6.0, 6.0 to 7.0, and 7.0 to 8.0
a n d three site classifications 0, 1, and 2, we then least-squares fitted the best estimate o f
l o g l o Sa o. T a b l e 5 a n d Figure 6 s u m m a r i z e the results o b t a i n e d in this way. As m a y be
seen f r o m F i g u r e 6, the scaling factor log1 o Sdo decreases with m a g n i t u d e in a way that
emphasizes the progressively smaller influence o f larger r e p o r t e d magnitudes on the
a m p l i t u d e o f seismoscope response. This is in agreement with o u r previous w o r k
(Trifunac, 1973) which suggested that close to the source o f energy release the stress
d r o p a n d the final D C c o m p o n e n t o f g r o u n d displacement should govern the overall

TABLE 4"
LOGARITHMS'~OF THE AMPLITUDESAo IN MILLIMETERS,WITH WHICH A
STANDARD TORSION SEISMOMETER (To = 0.8, V = 2800, h = 0.8)
SHOULD REGISTERAN EARTHQUAKEOF MAGNITUDEZERO

A A A
(km) - log Ao (km) --log Ao (km) - log Ao

0 1.4 150 3.3 390 4.4

5 1.4 160 3.3 400 4.5
10 1.5 170 3.4 410 4.5
15 1.6 180 3.4 420 4.5
20 1.7 190 3.5 430 4.6
25 1.9 200 3.5 440 4.6
30 2.1 210 3.6 450 4.6
35 2.3 220 3.65 460 4.6
40 2.4 230 3.7 470 4.7
45 2.5 240 3.7 480 4.7
50 2.6 250 3.8 490 4.7
55 2.7 260 3.8 500 4.7
60 2.8 270 3.9 510 4.8
65 2.8 280 3.9 520 4.8
70 2.8 290 4.0 530 4.8
80 2.9 300 4.0 540 4.8
85 2.9 310 4.1 550 4.8
90 3.0 320 4.1 560 4.9
95 3.0 330 4.2 570 4.9
100 3.0 340 4.2 580 4.9
110 3.1 350 4.3 590 4.9
120 3.1 360 4.3 600 4.9
130 3.2 370 4.3
140 3.2 380 4.4

*Reproduced from Richter (1958).

tSince A o is less than 1, its logarithm is negative, and the table shows
values for - log A o.

s h a p e o f the response spectra, while the influence o f m a g n i t u d e progressively diminishes.

A s seen in F i g u r e 6, the slope defined by the two averages o f log1 o Sdo for 0 site classifica-
t i o n a n d for m a g n i t u d e intervals 6.0 to 7.0 a n d 7.0 to 8.0 is a l m o s t equal to - 1 , indicat-
i n g t h a t the p e a k seismoscope response a n d therefore the response spectrum a m p l i t u d e s
a r e practically i n d e p e n d e n t o f m a g n i t u d e in the m a g n i t u d e range larger than a b o u t 6. A
m o r e precise description o f the dependence o f log1 o Sao on the m a g n i t u d e scale a n d the
d e t e r m i n a t i o n o f the m a g n i t u d e for which the slope o f log 10 Sao ceases to be m a g n i t u d e -
d e p e n d e n t will, o f course, have to await a m o r e a b u n d a n t set o f accelerograph data,
covering a m u c h larger m a g n i t u d e range a n d recorded on a m o r e c o m p l e t e s a m p l e o f
 CORRELATION OF SEISMOSCOPE RESPONSE WITH M AND IMM 317

different r e c o r d i n g sites than the d a t a we have at o u r disposal now. In the meantime, we

h o p e that the trends indicated in F i g u r e 6 will prove useful for the p r e l i m i n a r y correla-
tions o f seismoscope d a t a a n d r e p o r t e d e a r t h q u a k e magnitudes, as well as for the
a p p r o x i m a t e predictions o f e a r t h q u a k e m a g n i t u d e where only seismoscope records are
available.
Recent source m e c h a n i s m studies o f strong g r o u n d m o t i o n close to the e a r t h q u a k e
energy release (e.g., Trifunac, 1972a; 1972b), as well as n u m e r o u s teleseismic observa-
tions o f b o d y - w a v e spectra (e.g., H a n k s and Wyss, 1972), have d e m o n s t r a t e d that large

TABLE 5
DEPENDENCE OF LOG.to(SaD) ON MAGNITUDE AND SITE CONDITIONS

Cr
One Standard
Magnitude Site log~o (SAD) Deviation No.
Range Classification* (era) logj o (Sdo) of Data

4-5 0 - 3.028 0.242 4

4-5 l - 3.339 0.296 3

5-6 0 - 3.353 0.460 23

5-6 1 - 3.470 0.328 15
5-6 2 - 3.739 0.408 3

6-7 0 - 3.651 0.308 83

6-7 1 - 3.699 0.291 33
6-7 2 - 3.938 0.469 11

7-8~ 0 - 4.455 0.248 7

Total 182++

*See Trifunac and Brady (1975).

tFor most earthquakes used in this work, the magnitude has been
calculated from the Wood-Anderson seismograph records and repre-
sents, by definition, the Local Richter magnitude ML. For earthquakes
larger than about 6.5 to 7.0, local stations usually go off scale and the
"magnitude" has to be determined from other teleseismic seismograms.
For this reason, there may be systematic deviations between the magni-
tude estimates for shocks less than and greater than 6.5 to 7.0.
~The four calculated seismoscope records not included in this table
either had the assigned magnitude less than 4.0 or no magnitude infor-
mation was available.

e a r t h q u a k e s lead to progressively greater energy content in the l o n g - p e r i o d waves, while

the high-frequency end o f the displacement spectrum falls off like co-2 to co-3. The
intuitive physical basis for such scaling has been given by Brune (1970) who p r o p o s e d an
a p p r o x i m a t e shape o f e a r t h q u a k e displacement spectra based on the co- 2 falloff at high
frequencies a n d with a flat D C displacement spectrum a m p l i t u d e p r o p o r t i o n a l to the
seismic m o m e n t . In this theory the corner frequency, where the flat D C spectrum ampli-
tude begins to fall off like co-2, decreases with increasing seismic m o m e n t a n d / o r the size
o f an earthquake. Thus, when the p r e d o m i n e n t frequency b a n d o f the recording instru-
m e n t is at frequencies smaller than the corner frequency, the m a g n i t u d e derived from the
peak o f a s e i s m o g r a m m a y be expected to be representative o f the e a r t h q u a k e " m a g n i -
t u d e " ; while if the frequency b a n d is higher than the corner frequency, the p e a k on a
318 M. D. TRIFUNAC AND A. G. BRADY

seismogram becomes more representative of the amplitude of the o9-2 part of the dis-
placement spectrum.
The Wood-Anderson seismometer, which represents the standard instrument for pub-
lished magnitudes for the majority of earthquakes studied in this paper, has a transfer
function proportional to f2 for frequencies shorter than 0.8 Hz and equal to unity for
frequencies higher than 0.8 Hz. Thus, for small earthquakes having high corner fre-
quency (> 1 Hz), the Wood-Anderson instrument can sample the amplitudes of the flat
portion of the displacement spectrum quite well. For large earthquakes, as the corner
frequency becomes smaller than about 1 Hz, the Wood-Anderson seismometer samples
the amplitudes of the high-frequency spectra and ceases to grow with magnitude (Brune,
1970).

-2

-3

?
-4
O
on
o
m

-5

-6
"F.-- D. 0.--¢. ~.--[. /'. - - 1:5

MAGNITUDE
Fie. 6. Dependenceof Iogto(Sdo) on magnitude and site classification(see Table 5).

When a seismoscope is used for magnitude determinations, the above qualitative

discussion should become even more applicable since a seismoscope samples only a
narrow part of the input displacemev.t spectrum close to 1.33 Hz. Therefore, it might be
expected that the scaling given by log1 o Sdo should be nearly constant for small magni-
tude earthquakes, whereas for large magnitudes it would decrease by one for each unit of
increasing magnitude, if one were to assume that the amplitude of the 09- 2 portion of the
displacement spectrum has some absolute maximum which is independent of earthquake
magnitude (Brune, 1970; Thatcher and Hanks, 1973).
The effect of site classification on loglo Sdo is consistent for all magnitude ranges
covered by the available data. Although not significantly different in all cases, the mean
values of log1 o Sdo are consistently smaller for "intermediate" and "hard" sites than for
the "soft" sites. These differences show that the seismoscope response recorded on "soft"
alluvium can be up to ~2.5 times larger than the corresponding response on "hard"
rock formations.
 CORRELATION OF SEISMOSCOPE RESPONSE WiTH M AND IMM 319

Rewriting equation (6)

Mseismoscope = logxo Sd(Cm)--log1 o A0(A)-log I o Sao (6)

and using Tables 4 and 5, one can compute the Ms~ismoscoo~ magnitude by observing the
maximum deflection on seismoscope records, c~. . . . and by computing S a from equation
(2). If many seismoscope records are available, then one standard deviation of such an
estimate, obtained by averaging over all seismoscope records, may be on the average
0.35 magnitude units (Table 5), which is close to the accuracy with which the local
Richter magnitude, ML, can be determined (Richter, 1958).
The difficulty in this approach, however, arises from the fact that the log10 Sao is
magnitude-dependent and that several iterations may be required before one arrives at an
estimate of the Mseismoscope magnitude. For large earthquakes where loglo Sao versus
magnitude attains the slope of - 1 , i.e., the seismoscope response becomes insensitive to
earthquake magnitude, the above method of computing Mseismoscope breaks down and,
of course, should not be used. Precisely at what magnitude level this takes place cannot
be determined from the present set of data, and one can only guess that this is probably
in the range of magnitude 6 or 7.
If one does know the approximate magnitude range for the earthquake studied, the
standard error of each individual estimate of magnitude determined from a seismoscope
record would be about 0.35 magnitude units. If one does not know the range of possible
magnitude of an earthquake and one works 0nly with Tables 4 and 5 using the trial and
error procedure, the results may not be unique, i.e., there may be two magnitude esti-
mates for the two adjacent magnitude intervals given in Table 5 and Figure 6. In that
case, the error of an estimate can be as large as one magnitude unit. In normal circum-
stances, however, at least several seismoscope records will be available, and by averaging
the results from all seismoscopes the accuracy of the determined magnitude may be
significantly improved.
The important assumption in deriving equation (6) and subsequently the average
values of log1 o Sdo (Table 5) has been that the apparent attenuation with distance for the
Western United States can be adequately described by the empirical data in Table 4.
These data have been tabulated by Richter (1958) and result from numerous calcula-
tions of the local magnitude scale. Therefore, our results in Table 5 merely represent an
extension of the methods used by Richter in deriving the definition of the local magnitude
scale. The magnitude-dependent changes of log~ 0 Sao (Table 5 and Figure 6) then reflect
the inferred variations between our estimate of the local magnitude in terms of Mseismoscope
and the "magnitude" reported in the literature for the 57 earthquakes used in our
analysis. For the majority of shocks used in our work, the reported magnitude does
represent the local Richter magnitude. For larger earthquakes, however, when the local
Wood-Anderson instruments go off scale, the magnitude reported in the literature and
used in our correlations represents the surface-wave magnitude. Table 2 of our paper
dealing with the correlation of seismic intensity scales with the peaks of recorded ground
motion (Trifunac and Brady, 1975) gives all pertinent data for the 57 earthquakes studied
in this paper and lists their assigned magnitudes.

CONCLUSIONS

In this paper, we presented correlations of peak seismoscope response Sd with the

Modified Mercalli intensity. Although the range of the seismoscope response has been
found to vary appreciably for a given intensity level, IMM, illustrating the imprecise
nature of scaling the level of strong shaking with the Modified Mercalli intensity, there is
320 M. D. TRIFUNAC AND A. G. BRADY

a definite trend of the mean seismoscope response which can be approximated by

1 100.2 88IMM,
s d ~.~

for IMM < VIII. In the absence of reliable estimates of intensity of strong ground shak-
ing, when seismoscope records are available, this correlation can prove useful in supply-
ing additional intensity information. Since the seismoscope represents a typical building
vibrating at its first mode with the period of 0.75 sec and fraction of critical damping
equal to 10 per cent, the above correlation also provides means for predicting the
expected amplitude of the relative displacement response and the pseudo-relative
velocity spectrum.
By subdividing all available data into three groups, corresponding to the type of
geological formations underlying the recording stations used in this study, we found no
systematic trends in the above correlation that would indicate strong dependence on site
conditions.
We extended the standard methods used for calculation of the local earthquake
magnitude, ML, and applied them to the calculation of the local magnitude, Mseismoscope,
using the recorded peak of the seismoscope relative displacement response S a. We found
that the scaling function that relates the magnitude computed from the seismoscope
reading, Sa, with the magnitude reported in the literature is strongly magnitude-depen-
dent. We found that the data used in this study is not adequate to describe this magni-
tude-dependent scaling factor over a sufficiently broad magnitude range of practical
interest. The available data indicate, however, that there exists an upper bound for Sa,
and therefore the relative displacement spectra, which is practically attained for an earth-
quake with magnitude somewhere between 6 and 7.

ACKNOWLEDGMENTS

We thank G. W. Housner, P. C. Jennings, J. E. Luco, and F. E. Udwadia for critical reading of the
manuscript and for offering valuable suggestions.
This research has been supported in part by the grants from the National Science Foundation and by
the Earthquake Research Affiliates Program at the California Institute of Technology.

REFERENCES
Brune, J. N. (1970). Tectonic stress and the spectra of seismic shear waves from earthquakes, J. Geophys.
Res. 75, 4997-5009.
Cloud, W. K. and D. E. Hudson (1961). A simplified instrument for recording strong motion earth-
quakes, Bull. Seism. Soc. Am. 51,159-174.
Fournier d'Albe, E. M. (1973). Report on proceedings of international symposium on strong earthquake
motion, Mexico City, August, 1972, Bull. Seism. Soc. Am. 63, 1857-1865.
Galitzin, B. (1913). Uber eine Dynamische Skala zur Sch~itzung von Makroseismischen Bewegungen,
St. Petersburg, Review by H. O. Wood in Bull. Seism. Soc. Am. 3, 90--93.
Hanks, T. C. and M. Wyss (1972). The use of body-wave spectra in the determination of seismic source
parameters, Bull. Seism. Soc. Am. 62, 561-590.
Hudson, D. E. (1958). The Wilmot Survey Type Strong-Motion Earthquake Recorder, Earthquake
Eng. Res. Lab., California Inst. of Tech., Pasadena.
Hudson, D. E. and W. K. Cloud (1967). An analysis of seismoscope data from the Parkfield earthquake
of June 27, 1966, Bull. Seism. Soc. Am. 57, 1143-1159.
Hudson, D. E. (1971). Strong Motion Instrumental Data on the San Fernando Earthquake of February
9, 1971, Earthquake Eng. Res. Lab., California }nst. of Tech. and Seismological Field Survey,
NOAA.
Kirkpatrick, P. (1927). Seismic measurements by the overthrow of columns, Bull. Seism. Soc. Am. 17,
95-109.
 CORRELATION OF SEISMOSCOPERESPONSE WITH M AND IMM 321

Krishna, J. and A. R. Chandrasekaran (1965). Structural response recorders, Proc. Worm Conf. Earth-
quake Eng., 3rd, New Zealand, 143-150.
Lomnitz, C. (1970). Major earthquakes and tsunamis in Chile during the period 1535 to 1955, Geol.
Rundshau 59, 938-960.
Medvedev, S. V. (1965). Elementary Seismology, Israel Program for Scientific Translations, Jerusalem.
Medvedev, S. V. and W. Sponheuer (1969). Scale of Seismic Intensity, Proc. World Conf. Earthquake
Eng., 4th, Santiago, Chile, A2, 143-153.
Richter, C. F. (1958), Elementary Seismology, Freeman, San Francisco.
Scott, R. F. (1973). The calculation of horizontal accelerations from seismoscope records, Bull. Seism.
Soc. Am. 63, 1637-1661.
Suyehiro, K. (1926). A seismic vibration analyzer and the records obtained therewith, Bull. Earthquake
Res. lnst., Tokyo Univ. 1+59.
Thatcher, W. and T. C. Hanks (1973). Source parameters of southern California earthquakes, J. Geophys.
Res. 78, 8547-8576.
Trifunac, M. D. and D. E. Hudson (1970). Analysis of station no. 2 seismoscope record--1966, Park-
field, California, earthquake, Bull. Seism. Soc. Am. 60, 785-794.
Trifunac, M. D. (1972a). Stress estimates for San Fernando, California, earthquake of February 9,
1971 : Main event and thirteen aftershocks, Bull. Seism. Soc. Am. 62,721-750.
Trifunac, M. D. (1972b). Tectonic stress and source mechanism of the Imperial Valley, California,
Earthquake of 1940, Bull. Seism. Soc. Am. 62, 1283-1302.
"l-'rifunac,M. D. (1973). Analysis of strong earthquake ground motion for prediction of response spectra,
Intern. J. Earthquake Eng. Struct. Dynamics 2, 59-69.
Trifunac, M. D. and A. G. Brady (1975). On the correlation of seismic intensity scales with the peaks of
recorded strong ground motion, Bull. Seism. Soc. Am. 65, 139-162.
Ulrich, F. P. (1941). Progress Report for 1939 of the Seismological Field Survey of the U.S. Coast and
Geodetic Survey, Bull. Seism. Soc. Am. 31, 107-119.

EARTHQUAKEENGINEERINGRESEARCHLABORATORY
CALIFORNIAINSTITUTEOF TECHNOLOGY
PASADENA,CALIFORNIA91109

Manuscript received September 30, 1974

View publication stats