Вы находитесь на странице: 1из 17

207

Tectonophysics, 93 (1983) 207-223


Elsevier Scientific

EMPIRICAL
MID-PLATE

OTTO

Publishing

Company,

Amsterdam

- Printed

in The Netherlands

MAGNITUDE
AND SPECTRAL SCALING
AND PLATE-MARGIN EARTHQUAKES

RELATIONS

FOR

W. NUITLI

Department of Earth and Atmospheric Sciences, Saint Louis University, St Louis, MO 63103 (U.S.A.)
(Received

September

23. 1982)

ABSTRACT

Nuttli,

O.W.,

1983 Empirical

earthquakes,

magnitude

and spectral

In: S.J. Duda and K. Aki (Editors),

scaling

relations

Quantification

for mid-plate

and plate-margin
Tectonophysics. 93:

of Earthquakes.

207-223.
Published
for mid-plate
develop

scaling

occurred

laws for earthquake

near the border

plate interiors

spectra.

The mid-plate

the m,,- M, relation

can be represented

of the two lines corresponds

period

is at

of the spectrum
because

A principal
earthquakes

the derived

is a measure

of the spectral

is a measure

of the spectrum

motion,

whereas
area.

The derived spectra


the stress drop
periods.

spectra

amplitude

for frequencies
magnitude

of the mid-plate

increases

as the moment

are flat at the long periods,

for

relations

plate-margin

is the fact that

earthquakes,

ground

however,

for strike-slip

the large
motion

is a good

are flat at the long periods,

increases.

For the large plate-margin

of moment
earthquakes

periods,

indicate

faults, such as the San Andreas

plate-margin

occurs,

and M,

of strong ground
measure

of fault

implying

earthquakes

that

the derived

and a slope of two at short


earthquakes.

that an MS = 8.7 event requires

a fault length of 6000 km, obviously


has a rupture

is more

and have a slope of two

for the larger plate-margin

length of only 60 km, for a fault width of 20 km. A plate-margin

fault width of 20 km would require

earthquakes

varies as the fourth power of the corner period,

for mid-plate

width of 20 km, an MS = 8.2 earthquake

lines, of slope one

for which the comer

for the same m b or Ms. Because m b

at which damaging

have a slope of one at intermediate

relations

magnitude

periods.

is not. The latter,

The stress drops are almost independent

fault rupture

in both continental

it follows that mb is a good estimator

earthquakes

Their seismic moment

The moment-magnitude

expected

Mb- MS relation

exhibit two corner

at very long periods,

a moment-derived

at the short periods.


spectra

1 sec. The

have a higher M, value than the mid-plate

rupture

used in the study

occurred

by two straight

to the body-wave

of the mb- M, and MS-M,

feature

earthquakes

earthquakes

M,,

and to

plate interiors.

earthquakes

and two. The intersection


complicated,

their inter-relationship

All of the plate-margin

of the Pacific Ocean.

and oceanic

For mid-plate

MS, and seismic moment,

mb, surface-wave magnitude,


are analyzed, to determine

values of body-wave magnitude


and plate-margin
earthquakes,

earthquake
impossible.

of M, = 8.7 and
Assuming

a fault

length of about 850 km, which is as large as can be


of California.

Subduction

zone earthquakes,

other hand, if they have fault widths as great as 200 km, can give rise to MS = 8.7 earthquakes.

on the

tNTRODUCTION

The first magnitude

scale. which was proposed

magnitude,

M,,

earthquakes

in other parts of the world, Gutenberg

surface-wave
Gutenberg
equations
given

for southern

magnitude,

California

A4,, and Gutenberg

and Richter attempted


so that the numerical

earthquake.

However,

by Richter

earthquakes.

( 1935). was the local

To assign

and Richter
constants

to

(1936) developed

f 1945a.b) a body-wave

to adjust the arbitrary

magnitudes
magnitude,

a
~2~.

in the ?I~ and MS

values of M,, mh and MS would be the same for a

earthquake

allows this objective to be achieved.


Gutenberg,
who developed empirical

spectra

do not scale in a manner

which

Later this fact was realized by Richter and


equations to relate the three types of magni-

tude to one another {see, e.g., Richter, 1958). However, seismologists and engineers
still often speak of a magnitude
of an earthquake,
without specifying the scale
being used. Many published
studies of strong ground-motion
from a mixture of the various magnitudes.
without correcting

attenuation
suffer
for their numerical

differences. For example, in Californiait


is common to call Richter magnitude
the
N, value for earthquakes
for which Mt. is approximately
less than 6.5. and the MS
value for the larger earthquakes.
But an M, of 6.5 is equivalent
to an M, of 7.5.
which introduces a discontinuity
into the relation. It would be desirable to reserve
the term
Francisco

Richter
magnitude
exclusively
for Mr. By this logic the 1906 San
earthquake had a Richter magnitude of 6.75-7 (Jennings and Kanamori,

1979). but an MS of 8.25 (Bolt, 1968).


Hanks and Kanamori (1979) pointed out that saturation of the mh scale occurs at
around 7 when 1-set period waves are used. They noted that mh prior to 1960 was
obtained
ably

by using longer

greater

than

7. They

period

wave amplitudes.

also noted

that

which gave mh values consider-

no California

earthquake

had an M,

greater

than 6.8 or an MS greater than 8.3. These saturation levels (about 7 for 1-set
all of these
~lr,.,and 8.5 for 20set MS) exist because for large enough earthquakes,
narrow-band
time domain amplitude measurements
no longer measure gross faulting
characteristics
but only limited conditions
on localized failure along crustal fault
zones

(Hanks and Kanamori,


1979). To overcome these saturation
problems they
which they empirically
related to the
proposed
a scale of moment
magnitude,
radiated seismic energy under the assumption
that the stress drop is constant for
shallow earthquakes.
This moment magnitude
is directly proportional
to the very
long period amplitude level of the spectrum. When it was empirically related to Mt,,
mh and Ms. by Hanks and Kanamori (t979), the assumption
was made that stress
drop is independent
of seismic moment. It will be shown later that this assumption
of constant stress drop is valid for most California earthquakes,
which Hanks and
Kanamori (1979) were considering, but is not valid for mid-plate earthquakes. Thus
the best magnitude scales for estimating near-field strong ground motion remain M,~
and mb, because they are a direct measure of the amplitude of the ground motion at
periods near 1 sec. The moment magnitude, on the other hand. is a measure of the

209

fault rupture

area and the seismic moment,

some geophysical
tude

problems,

can be related

Hanks

to short

and Kanamori

observed

gross source properties

but not for estimating


period

ground

strong

motion

(1979) only if the average

motion.

of interest
Moment

in the manner

for

magni-

described

stress drop is constant

by

over the

range of seismic moments.

The data to be presented


works of many

in this paper

investigators.

some earthquakes

They consist

all three parameters

are empirical,

taken

from the published

of pairs of mb, Ms and IV, values.

have been determined,

resulting

For

in three pairs

of values. For other earthquakes only two of the parameters are available, and thus
just one of three possible pairs. The data are broken down into two basic types,
plate-margin

earthquakes

been restricted
latter

include

were data
Himalayas,

and mid-plate

to earthquakes
both

occurring

mid-continental

earthquakes.

In this study the former have

near the margin

and mid-oceanic

of the Pacific Ocean.

plate

earthquakes.

The

Not used

from earthquakes
that sometimes are called intra-plate,
such as in the
China, and the Colorado Plateau and Basin and Range regions of the

western United States, which probably


plate-margin
and mid-plate earthquakes.

are intermediate

in character

between

true

m b - Ms RELATIONS

Figure 1 shows the relation between mb, obtained from


Ms, obtained
from 20-set period waves, for mid-plate
symbols indicate the paper from which the data are taken,
solid-line curve is not a fit to the data, in the sense of fitting

I-set period waves, and


earthquakes.
The letter
as given in Table I. The
a mathematical
function

by least squares or some similar method, but rather the relation between mb and MS
which is obtained from the derived spectrum scaling relations. That is, the trends of
the m,-MS,

TABLE

data are used to generate

a set of spectra, and these

References
Letter

and MS-M,

m,-MO

to letter symbols

used in figures

Reference

Letter

Stein (1978)

Street and Turcotte

Stein and Okal

(1978)

Hasegawa

Wang et al. (1979)


Richardson

and Solomon

House and Boatwright


Nuttli and Kim (1975)
( 1979)

Stem ( 1979)

(1977)

and Wetmiller

(1980)

Geller ( 1976)
( 1977)

Nuttli et al. (1979)

Herrmann

Reference

(1980)

Liu and Kanamori

(1980)

Kanamori

and Anderson

Kanamori(

1977)

Street et al. (1975)


Herrmann

et al. (1981)

Singh et al. (1981)

(1975)

,
9-

MID-PLATE

EARTHQUAKES

8-

1
-I

7-

6MS

.
5-

3-

2-

II
Ol

mb

Fig. I. mb- Ms relation


scaling

relation

spectra

for mid-plate

earthquake>.

The solid-line

is obtained

from

the spectral

shown in Fig. 7.

in turn

are employed

to obtain

the curves

shown

in Fig. 1 and in similar

figures.
The curve of Fig. 1 consists

essentially

of two straight-line

segments.

For magni-

tudes less than mb = 4.5, the slope of the curve is unity, indicating
that the comer
periods of the corresponding
spectra are less than one second. For mb < 3, the data
points fall below the curve. The m,, and MS values of these small earthquakes
obtained from spectral amplitudes, and possibly are in error.
For 4.5 5 mh s 7.0, the slope of the m,-MS curve for mid-plate earthquakes
This can be satisfied
the square

by having

of the frequency,

the short-period

portion

of the spectrum

such as in Akis (1967) w model,

were
is 2.

fall off as

in Brunes

(1970)

model, and in the model of Hanks (1979). For mh > 7, which is near the upper limit
of observed mb values, the curve of Fig. 1 has a decreasing slope, corresponding
to
spectral comer periods greater than 20 sec. There are no observational
data to
support this predicted phenomenon.
Figure 2 presents
the m,-MS
data for plate-margin
earthquakes.
The first
impression is that there is much more scatter than in Fig. 1. This might be expected,
for plate-margin
earthquakes
frequently
involve complex rupture processes which
differ from event to event. A part of the curve is dashed to indicate that the
earthquakes
of this size which were used in the study did not have seismic moment
determinations.
The segment of the curve corresponding
to 4.2 -C mh 5 5.1 has a

211

$1

6_

PLATE

MARGIN

I1

EARTHQUAKES

?-

6-

Fig. 2. mb- Ms relation


the spectral

for plate-margin

scaling relation

earthquakes.

The solid-line

shown in Fig. 8. The dashed

values for which no seismic moment

portion

of the curve is obtained

line part of the curve corresponds

from

to magnitude

data were available.

slope of unity. The portion between 5.1 5 mb g 5.4 has a slope of about 4, and the
slope of the portion between 5.4 5 mb K 6 is about 2.5.
For the large magnitude earthquakes most of the data points of Fig. 2 lie to the
right of the curve. For these mag~tudes the curve may be looked upon as a
boundary or limiting relation, rather than an average-value one. The two most
anomalous points, which have the symbol e and are of mb 6.2, are for the 1965
Washington earthquake and the 1971 San Fernando, California earthquake, which
more properly should be classified as intra-plate rather than plate-margin earthquakes. Earthquakes involving movement on the San Andreas Fault would be called
plate-margin earthquakes.
MAGNITUDE-MOMENT

RELATIONS

Figure 3 presents the m,--MO data for mid-plate earthquakes. The curve, with two
linear segments of slope 1 and 2, is in reasonable agreement with the data except at
small magnitudes. The data of this figure, taken together with those of Figs. 1 and 5,

1
6
Fig. 3. m b - M, relation
scaling relation

for mid-plate

earthquakes.

The solid-line

curve is obtained

from

the spectral

shown in Fig. 7.

PLATE

1030-

I
MARGIN

EARTHQUAKES

1029 -

1028 -

E
k lot7
J
f

lo=-

to=-

10243

41

5I

6I

71

8I

mbob+ 0.4
Fig. 4. m,-h4,
P-wave
solid-line

relation

amplitudes
portion

for plate-margin

to account

earthquakes,

for anomalous

of the curve is obtained

with 0.4 added

absorption

from the spectral

to mb obtained

in asthenosphere
scaling relation

below

from teleseismic
the epicenter.

shown in Fig. 8.

The

02f-----MID-PLATE

EARTHWAKEs

101

IO2

IO

102

IO
:

Fig. 5. MS-M,
scaling

relation

shown in Fig. 7.

I
4

Fig. 6. Ms- M,, relation


scaling

for mid-plate

relation

relation

I
5

I
6

for plate-margin

shown in Fig. 8.

earthquakes.

The solid-line

earthquakes.

The solid-line

curve is obtained

curve is obtained

from the spectral

from the spectral

214

suggest

that

mb

has been

overestimated

for the small

earthquakes.

Herrmann (1976) showed that nz,, could be overestimated


of 102 dyne-cm and less.
Figure 4 shows the m,-MO data for plate-margin
with the curves obtained
units

to the observed

values

of California

amplitudes
absorption

from the spectral


mb values.

earthquakes

Chung

scaling
and

obtained

Street

earthquakes.

In order to agree

law, it was necessary

to add 0.4

( 1980) showed

Bernreuter

from

and

by 0.5 units for M,, values

I-set

period

that

teleseismic

nrh

P-wave

are underestimated
by approximately
0.4 units due to anomalously
large
in the asthenosphere
beneath the epicenter. Presumably
this occurs for

most shallow plate-margin

earthquakes.

although

the correction

for all of them may

not be exactly 0.4 units.


The MS-M,, relation for mid-plate earthquakes
curve of unit slope indicates there is a one-to-one
orders

of magnitude.

MS = 8.5, where

The spectra

the corner

predict

period

is given in Fig. 5. The straight-line


relation between MS and M, for 6

that the slope should

exceeds

20 sec. There

increase

are no data

at about

to test this

expected behavior.
The MS-MO data for plate-margin
earthquakes,as
given in Fig. 6, show reasonable
agreement with the relation predicted by the spectra. This figure further suggests
that some of the large scatter in the mb-MS relation
in accurately
SPECTRAL

determining

for plate-margin

m,,

of Fig. 2 results from difficulty

earthquakes.

SCALING RELATIONS

Figure 7 presents the spectral scaling relation for mid-plate


earthquakes,
as
derived from the empirical data of Figs. 1. 3 and 5. It is characterized
by a flat or
level portion at long periods. by a segment of slope 2 at short periods, and by a
corner

frequency

w which

satisfies

Mow4 = constant.

A similar

moment-corner

frequency relation had been proposed by Street et al. (1975). Street and Turcotte
(1977) and Herrmann
and Nuttli (1980) from spectral data of earthquakes
in the
central and eastern United States. This relation implies that stress drop increases as
seismic moment increases,
is independent
of moment

which differs from the usual assumption


that stress drop
(Aki, 1967; Hanks and Kanamori.
1979) or that Mow3 =

constant.
Figure 7 was derived by assuming that the I-set spectral amplitude is directly
to Ms.
proportional
to mb, and the 20-set spectral amplitude directly proportional
By trial and error the corner period was shifted back and forth until the Mb-MS,
m,-MO and MS-MO relations thus obtained appeared to give a satisfactory fit to the
data of Figs. 1, 3 and 5. Subjective

judgment

was used to evaluate

the quality

of the

fit, rather than mathematical


curve-fitting.
The upper curve of Fig. 7 corresponds
to an m,, of 7.2 and an MS of 8.7, which
are taken as likely upper limits for these scales on the basis of historical earthquake
information.
However, the spectra themselves do not place such an upper limit on

215

T bed
Fig. 7. Derived
fourth

spectral

scaling

power of the comer

relation

period,

these mb and MS values,


particular

for mid-plate

as indicated

earthquakes.

by the dashed

i.e., there is no saturation

form of spectral

The seismic

moment

varies

as the

line.

of either scale implied

by this

scaling.

The relation Moo4 = constant and the spectral slope of 2 at the shorter periods
leads to MS increasing twice as fast as Mu for mb > 4.5, where 4.5 is the mb value at
which the comer

period

is 1 sec. The data of Fig. 1 require

that such a relation

be

satisfied.
Much more difficulty was experienced in attempting to derive a set of spectra for
plate-margin
earthquakes
that satisfy the data of Figs. 2, 4 and 6, because the data
do not seem to conform to such simple relations as observed for mid-plate earthquakes and because there is a large amount of scatter in the data. Figure 8 presents a
scaling relation for plate-margin
earthquakes,
obtained
by trial-and-error
in an
attempt to satisfy the data of Figs. 2, 4 and 6. This spectral scaling relation may be
considered
to be an extreme or limiting case, as probably
also is the mid-plate
spectral scaling relation of Fig. 7, with spectra of many earthquakes
lying between
the two.
A distinctive

feature of the spectra of Fig. 8 is the presence

of a segment

of slope

1030

T,,

llll!,

PLATE

,,>,,

MARGIN

,llil,

EARTHQUAKES

,/

.,,,

102g-

102e-

102?--

E
%

1026_

I
P
I

I
iO23-

, 6,

102

5-

~~

lo

us

.,,,,,,,,,,i

IO

0.1

loo

1000

T (set)
Fig. 8. Derived spectral
0.4 units to account

scaling relation

for absorption

for plate-margin

earthquakes.

The mb value is corrected

by adding

in the asthenosphere.

equal to 1, in addition to the flat part at the long periods and the part of slope 2 at
the short periods. Thus there are two corner periods. For the longer of the two
corner periods

the relation

Mow4 = constant

applies

for M, B 1O28 dyne-cm,

but this

relation breaks down for smaller moments, as can be seen by the dashed line in the
figure. The lesser of the two corner periods appears to bear no simple relation to
seismic moment, except that it increases as the moment
paper presents no supporting data, for MO < 10 23 dyne-cm

increases. Although this


the spectra of Figs. 7 and

8 presumably
would be coincident.
Savage (1972) showed that the lesser of the two comer

periods

is proportional

to

the fault rupture width and the greater is proportional


to the fault length. In Figs. 7
and 8 the lesser comer periods are approximately
the same for a given mb, whereas
the greater comer period is larger for the plate-margin
spectra. This implies that the
rupture widths are similar for the two classes of earthquakes, but that plate-margin
earthquakes have relatively larger rupture lengths for a given ltrb value.
The mb values indicated in Fig. 8 actually are not the observed teleseismic P-wave
values, but rather values which are corrected for anomalous
absorption
of I-set

217

P-wave

energy

assumed

in the asthenosphere.

As discussed

to be 0.4 mb units for most plate-margin

for Fig. 4, this correction

is

earthquakes.

DISCUSSION

Figure
Gutenberg
relation

9 compares
and

the m,-MS

Richter

is commonly

relations

(Richter,

obtained

1958), namely

used, even though

with

that

the mb values were based on amplitudes

P waves of 5-10 set period, whereas present


of 1-set period. Thus it is not surprising

in this study

of

mb = 1.59 MS - 3.97. The latter

practice

uses the amplitude

that the Gutenberg-Richter

of

of P waves

mb values, for a

given MS, are larger. Except for MS values between 4 and 6, the m,-MS curves for
mid-plate and plate-margin
earthquakes are almost parallel, the separation being 0.4
mb units.
Figure 10 compares the m,-MO relations for mid-plate
quakes. For a given mb, (corrected for anelastic attenuation
plate-margin
earthquakes), M,, is larger for the plate-margin
illustrates

the difficulty

encountered
,

....-. GUTENBERG-RICHTER
--PLATE MARGIN EARTHQUAKES
MID-PLATE
EARTHQUAKES

in trying

and plate-margin
earthin the upper mantle for
earthquakes.

This figure

to assess a size to the two classes of

9-

4-

3-

2-

I-

Fig. 9. Comparison
from l-xc

of mb- Ms relations. The tnb of plate-margin

IO-set P-wave amplitudes.


value.

and mid-plate earthquakes

P-wave amplitudes, whereas the mb of the Gutenberg-Richter


The rnb value for plate-margin

earthquakes

is obtained

relation was obtained from S- to


is the observed,

or uncorrected,

Fig. 10. Comparison


plate-margin

of m b - M, relations

earthquakes

for plate-margin

is the value corrected

and mid-plate

for absorption

earthquakes.

in the upper

mantle,

The m h vatue. for


not the observed

value.

earthquakes, in the sense of making statements that one earthquake is greater than
another. If the amplitude of the high frequency, strong ground motion (and thus the
damage resulting from ground shaking) is of concern, then mb is the proper
parameter to use to compare mid-plate and plate-margin earthquakes. If, on the
other hand, fault rupture area is of concern, then MO is the parameter to use for
comparison. Thus, from the earthqu~e engineers point of view, m,, would be the
parameter of most value for quantifying earthquakes, whereas the structural geologist would prefer MO (or M, the moment magnitude).
Figure 11 compares Ms-M, relations for the two classes of earthquakes. For
mid-plate earthquakes Ms is in a one-to-one relation with MO over almost the emire
range of moments, whereas for plate-margin earthquakes the relation only is valid
for A4s< 6.5,
Figure 12 shows how the spectra relate to one another. For periods less than 10
set the two types of spectra do not differ greatly from each other. However, at longer
periods the plate-margin spectral amplitudes rise far above the mid-plate amplitudes.
One possible explanation of the intermediate slope of unity for the plate-margin

219

1;;
----

PLATE

MARGIN

EARTHQUAKES

Fig. 11. Comparison

of Ms- M, relations

for plate-margin

and mid-plate

earthquakes.

earthquakes is fractional stress drop (Brune, 1970). Another is a ratio of fault length
to fault width appreciably greater than one (Savage, 1972).
Seismic moment,
TABLE

M,, can be related to average stress drop, Au, and fault rupture

II

Source characteristics

Type

of mid-plate

MS

and plate-margin

earthquakes

Mo

ii

A2

A0

(dyne-cm)

(m)

(km2)

(km)

(km)

(bars)

8.7

4.0.102

10

1,200

60

20

230

8.5

2.5. lo*

1,100

55

20

170

7.5

2.5. 1O26

380

25

15

130

6.5

2.5. IO*

75

15

90

5.5

2.5. 1O24

0.3

25

50

4.5

2.5. 1O23

0.1

1.5

Plate-margin

8.2

4.0. 102*

(Strike-slip)

7.5

1.6. 102

2,400

Plate-margin

8.7

6.0. 1O29

15

120,000

(Subduction)

8.2

4.0.102s

17,000

170

Mid-plate

7.5
17,000

865

30

20

43

160

15

33

600

200

35

100

43

--

/
i
I

MID-PLATE

1029
I
,

1028

_a,

.I-*----7

#I
.f

_____-----

T (sac)
Fig. 12. Comparison of spectral scaling relations for plate-margin and mid-plate earthquakes. The mb
value for plate-margin earthquakes is the value corrected for absorption in the upper mantle, not the
observed value.

area, A, using (Aki, 19661.


M, = pAfi

and (Geller,

1976):

ho = (7/16)M,[m/(

LW)j3

where p is the rigidity modulus of the crustal rocks, taken to be 3.3 . 10 dynes/cm,
B is the average fault displacement,
L is fault rupture length and W = A/L is fault
rupture width. Combining
these equations with the data of Fig, 11, estimates of fault
rupture area and average stress drop can be made for mid-plate and plate-margin
earthquakes,
as shown in Table I.
In Table II the values of L> have been assumed. If smaller values were used, the
rupture area would be larger and the stress drop smaller. Also, the breakdown of the
area, A, into a length, L, and a width, W, is arbitrary,
but considered
to be
reasonable.

221

Table II offers an explanation as to how the 18 1l- 1812 New Madrid earthquakes
could have such relatively short rupture lengths and large surface-wave magnitudes.
Nuttli (1980) estimated the December 16, 1811 event to have an MS of 8.6, the
January 23, 1812 to have an Ms of 8.4, and the February 7, 1812 event to have an
MS of 8.7. He also observed that the December earthquake occurred on the southern
branch of the fault system, of length 125 km, that the January earthquake probably
occurred on the 60 km long central branch, and that the February earthquake
occurred on the 75 km long northern branch. A reasonable value for rupture width is
20 km, as the earthquakes had their foci in the crust and may have broken the
earths surface. Table II shows that such large magnitude earthquakes could occur in
a mid-plate en~ronment, and would not require large rupture lengths or areas. Table
II also shows that a mid-plate earthquake the size of the 1886 Charleston, South
Carolina event (MS of about 7.5) would have only a 25 km rupture length.
Table II indicates that the 1906 San Francisco earthquake of MS = 8.25 likely
represents the maximum earthquake for the San Andreas fault, because of the large
value of the rupture length. The rupture width cannot be much greater than 20 km
for this strike-slip, plate-margin earthquake. Even if the average fault displacement
were doubled, to 15 m, the rupture length would be 400 km. Hanks and Kanamori
(1979) concluded that California earthquakes of width 15-20 km must have an
upper seismic moment limit of approximately 1O28dyne-cm if fault lengths do not
exceed several hundred kilometers.
It has been observed that subduction zones can generate earthquakes of MS = 8.7,
e.g. the 1964 Alaska earthquake. If a 200 km rupture width is permitted, Table II
shows that the rupture length would be approximately 600 km. These values are
large, but within the realm of possibility.
Hanks (1979) noted that the logarithmic mean of the stress drops of California
earthquakes is approximately 30 bars, which is consistent with the stress drops given
for the plate-margin earthquakes in Table II. However, for mid-plate earthquakes
the stress drop increases with moment, as shown in the table. For smaller mid-plate
and plate-mar~n earthquakes than those considered in Table II, the average stress
drop likely would be in the range of I- 10 bars, as observed by Street and Turcotte
(1977) for eastern North America earthquakes.
CONCLUSIONS

Empirical Mu, MS and h4, interrelations suggest that the spectra of mid-plate and
plate-margin earthquakes scale in a different manner. The mid-plate spectra behave
in a way that indicates that the average stress drop increases with seismic moment,
whereas the plate-margin earthquakes have a nearly constant stress drop.
Although the seismic moment is directly proportional to the long period spectral
amplitude and to the fault rupture area, for mid-plate earthquakes it is not directly
related to the short-period portion of the spectrum, the part which is related to

272

damaging

strong ground

motion.

Thus for mid-plate

tude. M. is not a good measure


definition

of M is based on western

have seen, are not applicable


margin,

likely

of earthquakes

where

represent

between

United

extreme

those

presented

That

mid-plate

is. there

probably

and magnitude-moment

in this

magniThe

which, from what we

in this paper,

conditions.

a moment

of an earthquake.

environments.

considered

whose spectra

earthquakes

potential

States empiricisms,

for other tectonic

The two classes of earthquakes


number

of the damage

paper.

and plateare a large

relations

In particular,

lie some-

earthquakes

with

hypocenters
near plate margins but not associated with plate-margin
movement
itself can be expected to lie in this intermediate
category. However, only further
research

can determine

if this intuitive

reasoning

is correct.

ACKNOWLEDGMENTS

I wish to thank Dr. R.B. Herrmann for helpful discussion, and Dr. S.J. Duda for
inviting me to present this paper at the meeting of the International
Association of
Seismology and Physics of the Earths Interior in London, Canada. The research
reported in this paper was supported in part by the
Projects Agency, monitored
by the Air Force Office
Contract
F49620-79-C-0025,
and in part by NSF
Earthquake
Hazards Mitigation
Program, Division

Defense Advanced
Research
of Scientific Research, under
Grant PFR-7909795
of the
of Civil and Envirionmental

Engineering.
REFERENCES
Aki, K., 1966. Generation
Bull. Earthquake

and propagation

of G waves from the Niigata

earthquake

of June

16, 1964.

Res. Inst., Tokyo Univ., 44: 23-88.

Aki, K., 1967. Scaling law of seismic spectrum.

J. Geophys.

Res., 72: 1217-1231.

Bolt, B.A., 1968. The focus of the 1906 California

earthquake.

Brune, J.N., 1970. Tectonic

of seismic shear waves from earthquakes.

stress and the spectra

Bull. Seismol. Sot. Am.. 58: 457-472.


J. Geophys.

Res., 75: 4997-5009.


Chung.

D.H. and Bernreuter,

Lawrence

Livermore

D.C.. 1980. U.S. Nuclear

Laboratory

UCRL-52745,

Geller. R.J., 1976. Scaling relations

for earthquake

Regulatory

Commission

NUREG/CR-1457

and

41 pp.
source parameters

and magnitudes.

Bull. Seismol. Sot.

Am., 66: 1501-1523.


Gutenberg,

B., 1945a. Amplitudes

of P. PP and S and magnitudes

of shallow earthquakes.

Bull. Seismol.

Sot. Am.. 35: 57-69.


Gutenberg,

B., 1945b. Magnitude

determination

for deep-focus

earthquakes.

Bull. Seismol. Sot. Am., 35:

117-130.
Gutenberg,

B. and Richter,

C.F.,

1936. On seismic waves (third

paper).

Gerlands

Beitr. Geophys.,

47:

73-131.
Hanks.

T.C.. 1979. b values and wmy seismic source models:

along

active

Geophys.
Hanks,

crustal

fault

zones

and

the estimation

implications

for tectonic

of high-frequency

strong

structure
ground

variations
motion.

Res.. 48: 2235-2242.

T.C. and Kanamori,

H.. 1979. A moment

magnitude

scale. J. Geophys.

Res.. 84: 2348-2350.

J.

223

Hasegawa,

H.S.

and

seismo-tectonic
Herrmann,

R.B.,

tectonic

Wetmiller,

Herrmann,

of 19 August

1979 and

its

for eastern

North

American

earthquakes

with

Res., 84: 3543-3552.

O.W., 1980. Strong

7th World Conf. Earthquake

earthquake

Notes, 5 I : 23-37.

wave focal mechanism

J. Geophys.

R.B. and Nuttli,

1980. The Charlevoix

Earthquake

1979. Surface

implications.

Herrmann,

R.J.,

environment.

motion

Eng., Geoscience

investigations

Aspects,

in the central

Part II, Istanbul.

United

States. Proc.

pp. 533-536

R.B., Park, S.K. and Wang, C.Y., 1981. The Denver earthquakes

Bull. Seismol.

of 1967-1968.

Sot. Am., 71:731-745.


House,

L. and Boatwright,

seismic gap, Alaska.


Jennings,

P.C. and

records.

J., 1980. Investigation

J. Geophys.

Kanamori,

of two high stress drop earthquakes

in the Shumagin

Res., 85: 7151-7165.

H.,

1979. Determination

of local

magnitude,

M,,

from

seismoscope

Bull. Seismol. Sot. Am. 69: 1267-1288.

Kanamori,

H., 1977. The energy release in great earthquakes.

Kanamori,

H. and Anderson,

D.L.,

1975. Theoretical

J. Geophys.

Res., 82: 2981-2987.

basis of some empirical

relations

in seismology.

Bull. Seismol. Sot. Am., 65: 1073-1095.


Liu, H.L. and Kanamori,
the waveforms
Nuttli,

H., 1980. Determination

O.W., 1980. Seismic hazard

Int. Meet., Sot. Explor.


Nuttli,

of source parameters

of mid-plate

earthquakes

from

of body waves. Bull. Seismol. Sot. Am., 70: 1989-2004.


associated

Geophys.,

O.W. and Kim, S.G.,

with the New Madrid

faulted zone. Tech. Pap., 15th Annu.

6: 3469-3484.

1975. Surface-wave

magnitudes

of Eurasian

earthquakes

and explosions.

Bull. Seismol. Sot. Am., 65: 693-709.


Nuttli,

O.W.,

intensity

Bollinger

G.A.,

and body-wave

Richardson,

and Griffiths,

magnitude.

R.M. and Solomon,

D.W.,

SC.,

1977. Apparent

stress in the plates. Pure Appl. Geophys.,


Richter,

CF.,

1935. An instrumental

Richter,

CF.,

1958. Elementary

Savage,

J.C., 1972. Relation

along the Mexican

of corner

Stein,

subduction

S., 1978. An earthquake

Geophys.

J. R. Astron.

Stein, S., 1979. Intraplate


Geophys.
Stein,
Street.

R.L.

magnitudes

San Francisco,

to fault dimensions.

and tectonic

Cahf., 768 pp.


J. Geophys.

J., 1981. Seismic gaps and recurrence

Res., 71: 3788-3795.

periods

of large earthquakes

Bull. Seismol. Sot. Am., 71: 827-843.

on the Chagos-Laccadire

on bathymetric

features:

1978. Seismicity

and

R.B., 1976. Some problems


Earthquake

Turcotte,

F.T.,

and intensities.

United

and tectonics

plate. J. Geophys.

Ridge

and its tectonic

implications.

the 1968 Emperor

Trough

earthquake.

J.

of the Ninety

east Ridge

evidence

for

area:

Res., 83: 2233-2246.


with using magnitude

scales for eastern

North

Notes, 47 (3): 37-46.

1977. A study

of northeastern

North

American

spectral

moments,

Bull. Seismol. Sot. Am., 67: 599-614.

R.B. and Nuttli, O.W., 1975. Spectral

States earthquakes.

Geophys.

J. R. Astron.

S.C., Geller, R.J. and Stein, S., 1979. An intraplate

Geophys.

earthquakes

scale. Bull. Seismol. Sot. Am., 25: l-32.

Freeman,

frequency

swarm

of the Indian

earthquakes.

Street, R.L., Herrmann,


Wang,

Mercalli

Sot., 55: 577-588.

R.L. and Herrmann,

central

stress drop for intraplate

zone: a reexamination.

seismicity

E.A.,

deformation

American
Street,

modified

Res., 84: 4763-4768.

S. and Okal,

internal

between

115: 317-331.

earthquake

Seismology.

Singh, S.K., Astiz, L., and Havskov,

1979. On the relation

Bull. Seismol. Sot. Am., 69: 893-909.

Res., 84: 5627-5632.

characteristics

of the Lg wave generated

by

Sot., 41: 51-63.

thrust earthquake

in the South Chinese

Sea.

J.

Вам также может понравиться