Content uploaded by Allan Lindh
Author content
All content in this area was uploaded by Allan Lindh on Feb 15, 2016
Content may be subject to copyright.
THE 1906 SAN FRANCISCO EARTHQUAKE AND THE SEISMIC CYCLE
W. L. Ellsworth, A. G. Lindh, W. H. Prescott, and D. G. Herd
U.S. Geological Survey Menlo Park, California 94025
Abstract. The cyclic nature of strain accumu- some future time is a foregone conclusion
lation and release in great earthquakes on the (Allen, 1981). However, the timing of that
San Andreas fault appears to have modulated the earthquake and the specification of the sequence
historic pattern of seismicity in northern of events (if any) that will precede it are
coastal California. The region experienced many virtually unknown to us at the present time.
more strong earthquakes in the half century
preceding the 1906 earthquake than in the half
century following it. Fedotov and Mogi recog-
nized a similar seismic cycle accompanying great
earthquakes in the subduction zones of the
Western Pacific. The cycle consists of an
extended period of quiescence after a major
earthquake, followed by a period of increased
activity that leads to the cycle-controlling
earthquake and its foreshocks and aftershocks.
The timing of the next great earthquake cannot
be forecast with precision at present, although
it appears to be decades away. The reemergence
of M > 5 earthquakes since 1955 within the
latitudes of the 1906 surface rupture to the
north of San Jose, after an extended period of
quiescence following 1906, lead us to conclude
that the region is entering the active stage of
the cycle in which events as large as M 6, to
perhaps M 7 can be expected.
Introduction
On April 18, 1906 a M• 8¬ earthquake occurred
along a 430 km-long segment of the San Andreas
fault in central and northern California. The
ground breakage extended from San Juan Bautista
to at least as far north as Point Arena (Figure
1). Right-lateral surface offsets averaged 4 m
north of San Francisco and less then 2 m south
of San Francisco to San Juan Bautista (Figure
1). Changes in the angles between geodetic
monuments, measured after the earthquake, indi-
cated that the slip at the time of the earth-
quake occurred to a depth of about 10 km and
averaged more than 4« m north of San Francisco
and about 3 m south of San Francisco (Thatcher,
1975a).
This earthquake represents the most recent
event in a long sequence of very large earth-
quakes that have ruptured the San Andreas fault
in northern coastal California. That another
such earthquake will again rupture the fault at
It is the purpose of this paper to review the
observational evidence relating to the long-term
behavior of the San Andreas fault system along
the 1906 rupture and to relate these data to
simple models of the cyclic recurrence of great
earthquakes.
Framework for Prediction of Great Plate
Boundary Earthquakes
Work on the problem of the prediction of the
next 1906 earthquake is conducted in the face
of a null hypothesis that earthquake occurrence
is random in space and time. And clearly at
this time mout seismic activity can only be
described stochastically. Randomness, however,
is not a physical property of a process, but
rather reflects only our level of information
about that process. Our task then is to find
patterns, regularities and correlations that
refine our physical model of the earthquake
process.
The most significant progress to date in
localizing the occurrence of a future earth-
quake in space and time stems from the recog-
nition by Imamura that in a region where large
earthquakes are known to have occurred in the
past, the longer the time since the last earth-
quake, the greater the risk of another (Imamura,
1924; Aki, 1980). This concept has been formal-
ized into what has come to be known as seismic
gap theory by Fedotov (1965) and Mogi (1968).
They showed that along the Benioff zone beneath
Japan, the Kurile Islands, and the Kamchatka
Penisula, over a long period of time (several
hundred years), the aftershock zones of large
earthquakes entirely fill in the arc with very
little overlap. Repetition of large earthquakes
at any position along the arc is now understood
to be the direct result of the continuing motion
between the Eurasian and Pacific plates. Thus,
the average recurrence interval at a single site
is, to first order, the ratio of the average
126 ELLSWORTH ET AL.
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
Pt. Arena
Fort
Pt. Reyes,
San
122 ø 00'
..•
Santa Rosa
ß
Vacaville ß
4 0 ø 00'
38 ø 00'
Son Josa /-
Juan
Bautista
0 20 40 60 km
I i ß . , A i
1906 Rupture • 1906 Epicenter
FIG. 1. Geographic location map of central
and northern California. The average surface
displacement in the 1906 earthquake was
greater north of the epicenter than to its
south.
displacement in a single earthquake to the
average plate velocity at the site.
While the seismic gap theory has proven of
great value in predicting where large earth-
quakes will occur (McCann et al., 1978), irregu-
larity of the time intervals between successive
earthquakes at a given site is such that the
time of occurence cannot be confidently
predicted to within a decade. Another limita-
tion on its use in many areas, including
California, is a short recorded history of
earthquake occurrence that often leaves the
average interevent time and/or the date of the
last earthquake unknown.
The Seismic Cycle
An attempt to circumvent these difficulties
and better define the time of occurence of a
large earthquake on a known seismic gap was the
concept of a seismic cycle introduced by Fedotov
(1968). His analysis of the seismicity between
the recurrence of a large gap filling earthquake
suggested that once the aftershock activity had
subsided, a relative lull in seismic activity
was observed, followed by an increase in
activity to the time of the next event. Mogi
(this volume) has formalized this model and
characterized the activity during each period.
The cycle (Figure 2) begins with the seis-
mically quiet period (I) which occupies approx-
imately the first half of the interevent period.
Activity is low in the immediate source region
as well as in the surrounding area. The second
period (II) is characterized by increased
activity throughout the region. Activity in the
eventual source region may subside during the
later stages of this period (the "Mogi doughnut"
pattern). The third period (III) includes the
main event as well as its foreshocks and after-
shocks. As the aftershocks in the rupture zone
decay in frequency, regional activity may
increase through diffusion of the aftershock
zone. This leads into the quiescent period of
the next cycle.
STAGE *IT '111'
i
T/me
'm
T/me
i/
Time
FIG. 2. Schematic diagram of stages of the
seismic cycle. Hypothesized behavior of
earthquake rate (upper), regional stress
level (middle) and cumulative seismic
moment (lower) are illustrated.
ELLSWORTH ET AL. 127
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
Time-Predictable Model
A further refinement of the quasi-periodic
model of great earthquake occurrence has
recently been proposed by Shimazaki and Nakata
(1980). They found at three locations in Japan
that the time between successive great gap-
filling earthquakes was proportional to the
displacement in the first event. This "time-
predictable" model suggests that great earth-
quakes recur when the strain released in the
last event recovers. In principle, this allows
the estimation of the time of the future occur-
rence of a given great earthquake.
Such a calcuation could be made in several
different ways. If one could measure the slip
in the last event, then the local rate of plate
motion could be used to estimate the time of
occurrence of the next event (Sykes and
Quittmeyer, this volume). Similarly, if one
knew the strain drop in the last event, knowl-
edge of the post-earthquake strain recovery and
the current strain rate could also be used to
give such an estimate.
Recurrence Time of the Next Great
San Francisco Earthquake
Our current understanding of regional tec-
tonics suggests that the segment that broke in
the 1906 earthquake is presently locked and
accumulating elastic strain but that it has not
yet fully recovered the strain released in 1906
(Thatcher, 1975b). It is reasonable, then, to
estimate the recurrence time of a M 8 earth-
quake for this section of the fault from the
plate motion rate and the slip in the 1906
earthquake, using the time-predictable
hypothesis of Shimazaki and Nakata (1980).
Geologic Slip Rate
Plate tectonic calcuations indicate that the
long term slip rate between the North American
and Pacific plates averages 5« cm/yr (Minster
and Jordan, 1978). However, geologic recon-
structions of San Andreas fault offsets indicate
a long-term displacement rate of only about 2
cm/yr along the San Francisco peninsula and 3
cm/yr to the north (Herd, 1979). Most of the
remainder of the plate motion that is not
transmitted by the San Andreas fault appears to
be distributed along other strike slip faults
in the coast ranges (Figure 3). The long-term
slip rates for the San Andreas imply an average
recurrence time of roughly 150 years for a 3 m
event on the San Francisco peninsula segment of
the fault and 150 years for a 4« m event on the
segment of the fault north of San Francisco.
While these estimates of recurrence time coinci-
dentally agree, it should be noted that we do
not know that the next earthquake will neces-
sarily rupture both segments of the fault.
128 ELLSWORTH ET AL.
-38 ø
-$6 ø
0 /0 20 30 miles
0 I0 • • km
I I
FIG. 3. Principal active faults in the
gPeater San Francisco Bay region. Long-term
slip rate (cm/yr) and creep rate (parens)
appear next to fault name code. NSA-northern
San Andreas, PSA-peninsular San Andreas,
CSA-central San Andreas, SG-San Gregorio,
HI-Hosgri, CP-Calaveras-Paicines, ST-
Sargent, CS-Calaveras-Sunor, HD-Hayward,
CD-Concord, GV-Green Valley, RC-Rogers
Creek, MA-Maacama. Faults with slip rates
lower than 0.1 cm/yr omitted.
Bautista [
• • •ister
ß •.-•'Monte r %
Geodetic Strain Rate
Contemporary strain data are also consistent
with a roughly 150 year recurrence interval.
Strain rates across the San Andreas fault on
the southern San Francisco peninsula between
1970 and 1980 were about 0.60 microstrain/yr
(Prescott et al., 1980; Prescott, 1980). The
near-fault coseismic strain drop determined by
Thatcher (1975a) for this segment of the fault
was -115 microstrain, which gives a recurrence
time of about 190 years.
These calculations of recurrence time may be
significantly in error if the strain rate is
not constant in time. Thatcher (1975a, and
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
Table 2) suggested that the near fault strain
rate was accelerated for roughly 30 years
following the 1906 earthquake. If true, this
could significantly reduce the 190 year value
calculated above. However, the evidence for
the high post-earthquake strain rate comes from
the segment of the fault to the north of San
Francisco, where the coseismic strain drop was
substantially greater. A recurrence calcula-
tion for the Point Reyes region that includes
the effect of the rapid post-seismic strain
recovery gives a recurrence time of 220 + 40
years.
An important point that must be emphasized is
the uncertainty that must be attached to all
these estimates. The best case is for the San
Francisco peninsula where the two independent
estimates agree reasonably well. But, the
assumption that the next earthquake will occur
when the strain recovers or reaches a predeter-
mined threshold has not been demonstrated con-
clusively. The probability of a large earthquake
in the San Francisco Bay area would be much
more clearly defined if it could be established
that the seismic cycle described above applied
to large strike-slip earthquakes, and if it
could be established where in that cycle we are
at the present time.
Seismicity and Deformation Rates
In a discussion of the 1957 San Francisco
earthquake (M 5.3), Tocher (1959) anticipated
some of the key concepts in what was later
formalized as the seismic cycle hypothesis. In
that paper, he noted, as others had before
(Willis, 1924; Gutenberg and Richter, 1954),
that the seismicity rate in the San Francisco
Bay area was higher during the decades preceed-
ing the 1906 earthquake than in the 50 years
after. He further suggested that this quies-
cence ended with a sequence of moderate earth-
quakes (M 5«) that occurred in the mid-1950's.
Recently published research into the pre-
instrumental earthquake history of California
(Toppozada et al., 1979, 1980), taken together
with an additional quarter-century of instru-
mental data, permit us to re-examine Tocher's
suggestion, and compare these data with the
seismic cycle hypothesis. To that end, we have
assembled a catalog of M 5 earthquakes for
the region, covering the period from 1855
through 1980 (Appendix).
Regional Distribution of Earthquakes
Identification of meaningful variations in
seismicity data presumes the existence of
stable background patterns in those data. A
comparison of the historic pattern of moderate-
to-large earthquakes in the Coast Range with
microearthquakes from 1969-1980, shows that
spatial distribution has remained relatively
unchanged for at least the past 125 years
(Figure 4). The fault systems that produced
the strong earthquakes during the entire
historic period are the same ones that account
for contemporary background pattern of micro-
earthquakes, with the single exception of that
portion of the 1906 break north of San
Francisco. In that region very few events can
be associated with the San Andreas fault. In
fact, most of the events locate well to the
east of the fault with very few epicenters in
the 30-km-wide coastal zone between the San
Andreas fault and the Rodgers Creek - Maacama
fault zone or seaward of the San Andreas.
Within the San Francisco Bay region, the
overall distribution of events is similar to
that observed to the north, with most events
associated with strike-slip faults to the east
of San Francisco Bay. A few isolated zones of
epicenters are associated with the San Andreas,
including activity in the epicentral region of
the 1906 earthquake, to the west of San
Francisco. South of the junction of the San
Andreas and Calaveras-Paicines faults near San
Juan Bautista, microearthquake activity is
concentrated along the San Andreas fault, and
is distributed in a broad zone across the entire
breadth of the Coast Ranges.
Focal mechanism studies (Bolt et al., 1968;
Ellsworth and Marks, 1980; Olson et al., 1980)
show that throughout the entire region most
events have strike-slip focal mechanisms. Nodal
planes for right-lateral slip have strike direc-
tions that range from northwest to north, in
good agreement with the orientation and sense
of movement on geologically active faults. Some
thrust and normal faulting focal mechanisms are
observed, as are rare events that correspond to
left-lateral (reversed!) movement on north to
northwest striking faults.
Variations in Seismicity Rates
The historic record of moderate and large
earthquakes (Figure 5) suggests that there have
been significant temporal variations in the
earthquake occurrence rate over the past 125
years. The region within the latitudes of the
1906 surface rupture experienced many more
strong earthquakes in the half century preceding
the 1906 earthquake than in the half century
following it (Tocher, 1959; Kelleher and Savino,
1975; Turcotte, 1977). The fragmentary catalog
for the period from 1808 to 1854 (Table 1)
further suggests that the period of high
activity extends at least to the beginning of
the historic record. But the absence of any
earthquake reports in the first 32 years of the
records of missions in the San Francisco Bay
area ½1776-1807) might be taken as weak evidence
for a• earlier period of quiesence.
Since about 1955, the region has been more
active at the M 5 level than it was in the
preceding 50 years. Epicentral maps covering
the last five quarter-centuries clearly illus-
ELLSWORTH ET AL. 129
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
-eel
.i.I I
Ocr,
z •
o
ß ,.-I 0
• 0
ee
.,-I
'"- 0
.,-I
I
0 •
0
•ø
ß
ee
ß
• o (:x:)
130 ELLSWORTH ET AL.
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
CoN tlen#o½ioo
800.0-
600.0
z
Juan 8aulisto --
200.0-
o
o o o
o o
On no o
o o
% O • øøø
o
on
o 0 o
0 0 '
0
o o ø o•Ooo
0(•0 0 0
•o ø o oo • o Oo oo o
(•0 0 0 0 0
•oo • •o •
o oO o • o
• o o
• 0 • o o
• o o o 0o 0
o
0
O. O• 0 0
%•o ,.•o ,• ,.• ' ' , o ,
1900 I•
1920 1940 I•
YEARS A.D.
FIG. 5. Space-time plot of M>5 seismicity along the San Andreas fault system. Epicenters are
from Real et al. (1978) and Toppozada et al. (1979, 1980). Surface faulting in great earthquakes
of 1857 and 1906 shown by solid vertical lines. Surface faulting dotted offshore and dashed
where uncertain.
trate the change in the seismicity rate that
followed the 1906 earthquake and suggest the
return to a higher level of activity during the
past 25 years (Figure 6). Given the short
historic record, the apparent variations in the
earthquake rate must be subjected to some test
of their significance.
We have chosen to test the historic catalog
against the Poisson hypothesis of a constant
average event occurrence rate, t, for a popu-
lation of unrelated events using the Kolmogorov
test on the interevent times. According to the
Poisson model, the cumulative fraction of the
population with interevent times less than t is
F(t) = 1 - e
The maximum deviation between the observed dis-
tribution and F(t) is the calculated Kolmogorov
test statistic. When this test is applied to
the catalog covering the entire region, with
aftershocks removed, we cannot reject the
hypothesis that the distribution is Poissonian
at the 95% confidence level. But when we
restrict our attention to the portion of the
plate boundary that ruptured in the 1906 earth-
quake we reject the null hypothesis at the 95%
confidence level (Figure 7).
The non-Poissonian behavior of the catalog for
the region along the 1906 rupture is largely the
result of one very long interevent time. The
473-month interval from 1914 to 1955 greatly
exceeds the average interval of 44+79 months. If
we hypothesize that this long interval reflects
a change in the underlying event rate t,
possibly as a result of the strain drop of the
1906 earthquake, and recompute the Kolmogorov
test with it excluded, we find that the null
hypothesis cannot be rejected. Thus, we con-
clude that a first order feature of the data is
that event rate, A, precipitously declined
following the 1906 earthquake.
Deformation Data
Variations in the rate of earthquake produc-
tion and their possible association with the
1906 earthquake raise questions about the exis-
tence of concurrent variations in the rate or
distribution of strain accumulation. Relative
motion across the San Francisco Bay area is
distributed on at least four major faults
spanning a zone more than 80 km wide. Geologic
determinations of slip rate indicate that all
four transmit a part of the relative motion
between the Pacific and North American plates
(Figure 3). Offset geologic features indicate
a long term (millions of years) average slip
rate of 10 mm/yr on the San Gregorio fault
(Webber and Lajoie, 1977; Webber and Cotton,
1980) and 20 mm/yr on the San Andreas fault
(Cummings, 1968; Addicot, 1969). Uncertainties
in age and displacement are such that both of
these rates could vary by as much as 50%. Along
the Hayward and Calaveras-Sunol faults the
geologic rate is unknown, but south of San Jose
the geologic rate on the Calaveras-Paicines
fault is about 15 mm/yr and presumably this is
distributed between the Hayward and Calaveras-
Sunol faults (Herd, 1979).
Geodetic determination of the slip rate during
the period 1970 to 1980 generally agrees with
the geologic rate. Prescott et al. (1980;
Prescott, 1980) found rates of 10+3 mm/yr slip
on the San Andreas fault, 6+1 mm/yr on the
Hayward fault, and 6+1 mm/yr on the Calaveras-
Sunol fault. The slip associated with the San
Andreas fault does not appear at the surface,
ELLSWORTH ET AL. 131
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
TABLE 1. Strong Earthquakes of the Greater San Francisco Bay Region, 1836-1980.
Date Location 1 Magnitude 1 Comments Reference 1
1808/6/21 San Francisco? ? Destroyed all of the 2
buildings, in particular
barracks and other houses
of the Presidio of San
Francisco
1836/6/10 37.8 N 122.2øW 6.7 Ground breakage on the 3
Hayward fault
1838/6 37.6 122.4 ø 7.0 Probable ground breakage 3
on Peninsula San Andreas
1868/10/21 37.7 122.1 ø 6.7 Ground breakage on the 4
Hayward fault
1892/4/21 38.5 122.0 o 6.6
1906/4/18
Largest (?) in series of
earthquakes near the town
of Vacaville that included a
M 6.4 shock two days earlier
37.8 122.6 o5 8¬ 6 Felt foreshock assigned 4
epicenter off Golden Gate
by Reid (1910)
1911/7/1 37.25 121.75 ø ? 6.6 ? Intensity data indicate an 8
earthquake of comparable
size to M 5.9 earthquake
of 1979/8/6
1. All earthquakes are described in Townley and Allen, (1938). Locations and
magnitudes from 1836-1900 are from Toppozada et al., (1979, 1980).
2. Quoted from Gaceta de Mexico (892, 1808) by Orozco y Berra (1887)
3. Louderback, (1947).
4. Lawson et al. (1908).
5. Boore (1977).
6. Bolt (1968) demonstrates that an M value of 8¬ is consistent with
the teleseismic records. A moment-magnitude (Hanks and Kanamori, 1978) of 7.7 is
obtained from the seismic and geodetic moment determined by Thatcher (1975a).
7. This magnitude is listed in Gutenberg and Richter (1954). We
calcuate an intensity magnitude of 5.6 using relations of Toppozada (1975) and
intensity data from Templeton (1911).
8. Templeton (1911) and Wood (1912).
but rather is distributed across some 40 km
normal to the fault and presumably reflects
(elastic) strain accumulation. North of San
Francisco both geologic and geodetic determina-
tions of slip rate are less certain.
Table 2 summarizes the strain rate since 1906
for the geodetic nets shown in Figure 8. Because
of the spatial inhomogeneity of deformation in
the San Francisco Bay area, care is required to
avoid having spatial variations in strain rate
masquerade as temporal variations. For this
reason in Table 2 we have compared only the
strain rates obtained over different time
periods from a constant set of angles or
distances.
132 ELLSWORTH ET AL.
The first three cases, Primary Arc, Hayward
Net and CDWR Net (Table 2), compare geodetically
determined strain rates in the San Francisco
Bay area. Cases 1 and 2 were obtained from
angle observations. In neither case is the
change in strain rate (lines lc and 2c, Table
2) significant at the two standard deviation
level. Case 3 was derived from repeated measure-
ments of the length of 11 lines from the
California Department of Water Resources (CDWR).
The difference between the 1960-1967 and the
1970-1980 strain rate (line 3c, Table 2) is
apparently significant. Unfortunately a change
of procedures and instrumentation occurred in
1968-1969 and Savage (1975) found that there
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
1855 - 1879 1880 - 1904 1905-1929 1950-1954 1955 -1980
o
• o o
\
oo o •o
ß
o
M=5 o o o o o 0 o
6 0 [•oo
o
o
o o o o
FIG. 6. Seismicity of the San Francisco Bay region, M_>5, by quarter-century intervals from
1855-1980.
are systematic differences between the two
techniques. Thus the observed change is
probably not reliable. The fact that all three
cases indicate a decrease in strain rate may be
significant.
North of San Francisco Bay the data of both
Fort Ross and Point Reyes indicate a signifi-
cantly higher rate prior to about 1940 (lines
5c and 6c, Table 2; Thatcher, 1975a). The data
at Point Arena (line 4a, Table 2) are consistent
with the higher early rate, but the second time
period is too short to yield meaningful results.
South of the San Francisco Bay region, along
the central segment of the San Andreas fault,
the data indicate that plate motion is accom-
modated by rigid block motion across the San
Andreas fault (Savage and Burford, 1971;
Thatcher, 1979). Here, the geodetic data
z,., 0.6
0:2
• ' 0 5 i0 15
o• INTER EVENT TIME t (Years)
FIG. 7. Kolmogorov test of the interval time dis-
tribution of earthquakes in the San Francisco Bay
region with a Poisson model (smooth curve). Max-
imum deviation of observed distribution from
Poisson model, at vertical arrow, exceeds the ex-
pected range for a Poisson process at the 95% con-
fidence level.
clearly demonstrate that there have been no
gross changes in the slip rate since 1885
(Thatcher, 1979). Seismicity along this reach
of the fault similarly shows no significant
temporal variation either before or after 1906
(Figure 5). Apparently, this segment of the
fault is effectively buffered from the end-
effects of great earthquakes to the north.
To summarize the deformation data: north of
San Francisco the near-fault strain appears to
have been elevated for a period of 25 to 30
years following 1906; in the San Francisco Bay
area there is no hard evidence of any changes
in the rate of strain accumulation or slip, but
there is marginal evidence of a general decline
in strain rate with time. No data exist to
indicate whether the strain rates near the San
Andreas fault along the San Francisco peninsula
were high after 1906 as were the rates to the
north. South of the 1906 break, the strain
rate appears to have been constant since the
mid-1880's. Finally it is unlikely that any
significant increase in the rate of deformation
accompanied the increase in M 5 and greater
earthquakes that occurred about 1955.
Discussion
The hypothesis of a seismic cycle (Figure 2)
as advanced by Fedotov (1968) and Mogi (this
volume) is based principally upon observations
of the subduction zones of the Western Pacific.
While we must guard against premature acceptance
of its universality, the principal features of
the cycle do appear to be represented in the
historic record of seismicity discussed above.
Long-term variations in the production rate of
M 5 earthquakes in the San Francisco Bay region
indicate an active period (stage II) lasting
for at least 50 years before the 1906 earth-
quake, followed by a marked quieting of the
region for nearly 50 years after the earthquake
(stage I). If, as Tocher (1959) suggested, the
ELLSWORTH ET AL. 133
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
Table 2. Y is the direction across which right lateral shear is a maximum
and ¾is the rate of engineering shear strain across this direction
Period (ppm/yr) Source
Primary Arc
la 1907-1922 0.79 + 0.23 N 41 + 10
lb 1922-1948 0.46 + 0.17 N 36 + 13
lc Difference -0.33 + 0.29 -5 + 16
Hayward Net
2a 1951-1963 0.60 + 0.12 N 27 W + 6
2b 1963-1978 0.40 + 0.09 N 28 W + 7
2c Difference -0.20 + 0.15 1 + 9
CDWR Net
3a 1960-1967 0.87 + 0.10 N 24 W + 4
3b 1970-1980 0.42 + 0.02 N 26 W + 2
3c Difference -0.45 + 0.10 2 + 4
Pt Arena
4a 1906-1925
4b 1925-1930
4c Difference
2.28 + 0.48 N 72 W + 6
16.75 + 7.96 N 59 E + 14
Not meaningful
Fort Ross
5a 1906-1930 2.47 + 0.80
5b 1930-1969 0.33 + 1.07
Thatcher (1975b)
Thatcher (1975b)
Lisowski (unpublished)
Lisowski (unpublished)
This paper
This paper
Thatcher (1975a)
Thatcher (1975a)
N 57 W + 8 Thatcher (1975a)
N 49 W • 91 Thatcher (1975a)
5c Difference -2.14 + 1.34 -8 + 91
Pt Reyes
6a 1930-1938 2.27 + 0.74
6b 1938-1961 0.75 + 0.16
N 57 W + 10 Thatcher (1975a)
N 52 W • 9 Thatcher (1975a)
6c Difference -1.52 + 0.75 -5 + 13
quiet period following the earthquake ended in
the mid 1950's, then the region can expect the
continued occurrence of moderate earthquakes in
the coming decades, culminating in another great
earthquake. This may also imply that M 6-7
earthquakes (such as the 1836 and 1868 earth-
quakes on the Hayward Fault), capable of inflic-
ting serious damage and significant loss of
life, should be expected in the San Francisco
Bay area at the rate experienced in the 19th
century, or at an average rate of about one
each decade.
Our analysis of the seismicity of the San
Francisco Bay area leads us to the same general
conclusions as reached by Tocher. Seismicity
at the M 5 level since 1955 generally resembles
that from the 19th century in both their spatial
distribution and frequency, although the number
of events in the sample (nine) is too small to
conclusively establish the correspondence. The
134 ELLSWORTH ET AL.
available evidence, taken literally, suggests
that while the seismicty has increased, it has
not fully returned to the level seen before
1906 (Figure 9).
At present, the very general descriptive model
of a seismic cycle presented here is of little
predictive value as to the time of either large
earthquakes on any one of a number of faults, or
of the next great gap filling earthquake along
the 1906 break. The long duration of stage II
before the 1906 earthquake would, however,
suggest that the repeat of that earthquake is
not imminent. The potential for these ideas in
improving our estimate of the timing of the next
great earthquake lies in the development of a
quantitative physical model that predicts the
increase in seismic activity, while satisfying
the constraints provided by the geodetic
measurements of the strain field.
Certainly the simplest model to account for
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
M >5 A4
;500-
J
5.
200-
lee- •-•
e- i • , , , i , ,
1850 1900
1950 1980 2000
Years A.D.
FIG. 9. Cumulative count of earthquakes in the San Francisco Bay region by magnitude. Counts of
M 4 and M 3 earthquakes are incomplete before 1930 and 1942, respectively.
trolled by long-wavelength forces. The occur-
rence of great earthquakes on the San Andreas
fault, and steady, deep-seated movement between
the plates in the lower lithosphere modulates
the stress level acting on the system without
disturbing the details of the balance between
its parts.
Concerning what directions future work on
this problem might take, we feel compelled to
point out that the most significant conclusion
of this paper, that the Bay area may now be in
stage II of the seismic cycle, is also the
least certain. The seismic cycle concept itself
is yet but a working hypothesis, and the
significance of the increase in M 5 activity
has not been established beyond any doubt. Yet
for earthquake prediction and hazard reduction,
the potential implications of the cycle are
great. Previous estimates of the annual prob-
ability of M 6-7 earthquakes on individual
faults in the Bay area have been of the order
of 10-2. A cumulative estimate of 10-1 for
the region involves a probability gain of two
to five (Aki, this volume), if we are indeed in
stage II of the cycle.
The task now is to reduce the uncertainties
in the factors that enter into this calculation.
Detailed geologic work on the major faults in
the area might improve our knowledge of average
recurrence intervals and the time of the most
recent events. Geodetic experiments now in
progress are likely to provide a better model
of the strain recovery process, further refining
the geodetic estimates of recurrence time. Care-
ful monitoring of microearthquakes may help
refine our knowledge of the stresses acting at
depth on the fault, and might lead to identi-
fication of long-term and/or short-term precur-
sors to the next major earthquake.
Summary
1. Seismicity at the M 5 level was low in the
San Francisco Bay for the 50 years following
the 1906 earthquake, when compared to activity
during the previous 50 years.
2. Since about 1955, activity appears to have
increased, although it has not yet returned to
pre-1906 levels.
3. Offsets of geologic markers place
constraints on the long-term average slip rates
on a number of major Bay area faults. Faults
which accommodate a significant portion of the
North American-Pacific plate boundary motion
(5« cm/yr) include the San Gregorio (1 cm/yr),
San Andreas (2 cm/yr), Calaveras-Paicines (1.5
cm/yr), Hayward (3/4 cm/yr), Calaveras-Sunol
(3/4 cm/yr), Rodgers Creek (3/4 cm/yr), Maacama
(3/4 cm/yr) and Green Valley (3/4 cm/yr) faults.
4. 3 m of offset was observed on the San
Francisco peninsula section of the San Andreas
during the 1906 earthquake, and no significant
aseismic slip has been observed since then.
Combined with the long-term geologic rate on
this fault, this implies a long term average
recurrence time of 150 years for this section
of the San Andreas fault.
5. Geodetic data give an strain drop of 115
136 ELLSWORTH ET AL.
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
microstrain at the San Andreas fault on the San
Francisco peninsula during the 1906 earthquake.
Current measurements of the strain recovery in
the same region give 0.6 microstrain/yr,
implying an average recurrence time of 190
years.
6. Studies of the seismicity accompanying the
recurrence of great trench earthquakes in the
western Pacific have led to the concept of a
seismic cycle, the principal features of which
is quiesence following a great earthquake, with
increased activity during the second half of
the interevent period. Sesimicity data for the
San Francisco Bay area summarized above fit
this pattern, although they are not adequate to
confirm its applicability to strike slip earth-
quakes on this transform boundary.
7. If the seismic cycle is applicable and if
the estimates of 150-190 years for the average
recurrence interval are correct, then it is
likely that large (M 6-7) earthquakes are more
probable in the next 70 years than their absence
in the %ast 70 years might suggest.
8. However, even if all of these arguments are
correct, nothing is implied as to where or when
these earthquakes might occur, except that we
believe from the historic record and geologic
evidence that the faults named above are the
most likely candidates.
9. One can arrive at the same conclusion more
directly by simply observing that large earth-
quakes were far more common in the Bay area in
the 19th century than they have been in the
20th. They occurred at a rate of about one
event per decade in the 70 years prior to 1906.
10. Thus, we believe that a repeat of the 1906
earthquake on the San Andreas fault does not
seem likely in the next few decades, but that
one or more large (M 6-7) earthquakes on any
number of faults (including the segment of the
San Andreas fault on the San Francisco penin-
sula) are possible, or even likely, in the
coming decades.
Appendix. Historic Record of Damaging
Earthquakes
The brief pre-instrumental record of damaging
earthquakes in the coastal ranges of central
and northern California covers less than two
centuries and is strongly influenced by the
density and distribution of settlement in the
region. Even in the immediate San Francisco
Bay region the record does not approach an
acceptable level of completeness until after
the 1849 gold rush. Despite these difficul-
ties, several strong earthquakes from before
1849 are known, and reports suggest others
(Table 1). It should be stressed that the list
of events in Table 1 is fragmentary. No earth-
quakes were reported by the inhabitants of
either San Francisco and Mission Santa Clara
(near San Jose) during the first 32 years of
their establishment (1776-1807). No earth-
quakes are mentioned at all in the available
records from the Russian settlement at Fort
Ross (1812-1841). The absence of earthquake
reports in the records of these settlements
does not necessarily imply that there were no
severe earthquakes during this period, as the
records are principally concerned with economic
matters.
The quantity and quality of written records
of earthquakes dramatically improved with the
influx of settlers in the 1850's. Toppozada et
al. (1979, 1980) have systematically examined
19th-century newspapers and other sources in
compiling their catalog of pre-1900 seismicity.
We have adopted their catalog in our analysis
of this period. We estimate that the catalog
it is essentially complete down to M 5« after
1855 in the San Francisco Bay region. However,
the catalog contains only about half of the
events to be expected from the Gutenberg-
Richter frequency-magnitude law in the interval
from M 5 to M 5«. Data on earthquakes from
1900 onward are taken from several sources,
including Real et al. (1978); Bulletins of the
Seismographic Stations, University of
California, Berkeley; Savage and McNally
(1974); and U.S.G.S. Catalogs of Earthquakes
along the San Andreas Fault System. The
compiled catalog appears in Table A1.
No aftershocks of the 1906 San Francisco
earthquake are included in the catalog that we
assembled, unless the 1911 event on the
Calaveras-Paicines fault is considered as one.
This is because no M 5 aftershocks of the
earthquake have yet been documented, although,
Lawson et al. (1908) tabluated felt reports for
two years following the earthquake. Among the
reports that they list, those from Berkeley,
California were considered most reliable, due
to the efforts there of trained observers who
attempted to compile complete lists of felt
events. We have appended to their list the
listings of felt earthquakes at Berkeley for
the period 1910-1949 from Bolt and Miller
(1975), and have plotted the event rates versus
time to assess the impact on the catalog of
1906 aftershocks not being included (Figure
A1). The event rate decays in accordance with
Omori's law (t-l) until about 1910, and
appears to reach a constant value by about
1915. This suggests that the omission of 1906
aftershocks does not influence the conclusions
reached in this paper.
Earthquake magnitudes and epicentral coor-
dinates that we have used generally correspond
to those listed in the cited reference. The
only exceptions are for those events occurring
between 1900 and 1927 for which we have computed
local magnitudes from published felt area data
using relations given by Toppozada (1975). When
multiple sources listed different magnitude
values, we have averaged the available data.
Instrumental magnitudes determined from multiple
records (Savage and McNally, 1974) have been
ELLSWORTH ET AL. 137
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
TABLE A1. Magnitude 5 and Larger Earthquakes
in Coastal California from 36« øN to 39« øN,
1855-1980
Year Date Time Latitude Longitude Magnitude
1855 8 27 23:0 38N 12 122W 30 5.1
1856 1 2 0:0 37N 18 122W 12 5.3
1856 2 15 13:25 37N 36 122W 24 5.9
1858 11 26 8:35 37N 36 122W 0 5.9
1861 7 4 0:11 37N 42 121W 54 5.3
1864 2 26 8:40 36N 30 121W 30 5.8
1864 3 5 16:49 37N 24 121W 42 5.3
1864 5 21 2:1 38N 24 122W 18 5.2
1864 7 22 6:41 37N 30 121W 54 5.4
1865 3 8 14:30 38N 24 122W 36 5.3
1865 5 24 11:21 37N 6 121W 42 5.0
1865 10 8 20:46 37N 6 121W 54 6.2
1866 3 26 20:12 37N 6 121W 30 5.8
1866 7 15 6:30 37N 18 121W 12 5.7
1868 10 21 15:53 37N 42 122W 6 6.7
1869 10 8 9:30 39N 6 123W 6 5.2
1870 2 17 20:12 37N 12 122W 0 5.5
1870 4 2 19:48 37N 48 122W 12 5.1
1881 4 10 10:0 37N 18 121W 18 6.0
1882 3 6 21:45 36N 42 121W 12 5.5
1883 3 30 15:45 36N 54 121W 36 5.7
1884 3 26 0:40 37N 0 121W 12 5.3
1885 3 31 7:56 36N 42 121W 18 5.6
1885 4 2 15:25 37N 0 121W 24 5.5
1887 12 3 18:55 39N 18 123W 30 5.2
1889 5 19 11:10 38N 0 121W 54 5.6
1889 7 31 12:47 37N 48 122W 12 5.1
1890 4 24 11:36 36N 54 121W 36 5.9
1891 1 2 20:0 37N 12 121W 42 5.1
1891 10 12 6:28 38N 18 122W 18 5.6
1892 4 19 10:50 38N 30 122W 0 6.8
1892 4 21 17:43 38N 36 121W 54 6.3
1892 4 30 0:9 38N 24 121W 54 5.0
1892 11 13 12:45 36N 48 121W 48 5.2
1893 8 9 9:15 38N 24 122W 36 5.0
1897 6 20 20:14 36N 54 121W 24 6.0
1898 3 31 7:43 38N 12 122W 24 6.3
1898 4 15 7:7 39N 12 123W 48 6.7
1899 4 30 22:41 36N 54 121W 36 5.7
1899 6 2 7:19 37N 48 122W 36 5.5
1899 7 6 20:10 36N 42 121W 18 5.5
1902 5 19 18:31 38N 18 121W 54 5.5
1903 6 11 13:12 37N 36 121W 48 5.6
1903 8 3 6:49 37N 18 121W 48 5.6
1906 4 18 13:12 37N 42 122W 30 8.3
1910 3 11 6:52 36N 54 121W 48 5.7
1910 12 31 12:11 36N 50 121W 25 5.0
1911 7 1 22:0 37N 15 121W 45 6.0
1914 11 9 2:31 37N 10 122W 0 5.5
1916 8 6 19:38 36N 40 121W 15 5.4
1926 7 25 17:57 36N 36 120W 48 5.2
1926 10 22 12:35 36N 37 122W 20 6.2
1926 10 22 13:35 36N 34 122W 20 6.2
1926 10 24 22:51 37N 1 122W 12 5.6
1927 2 15 23:54 36N 57 122W 15 5.6
1939 6 24 13:2 36N 48 121W 33 5.2
1945 1 7 22:25 36N 46 121W 13 5.0
1945 8 27 9:13 37N 23 121W 43 5.0
138 ELLSWORTH ET AL.
1949 3 9 12:28 37N 2 121W 29 5.2
1951 7 29 10:53 36N 37 121W 14 5.0
1954 4 25 20:33 36N 55 121W 42 5.3
1955 9 5 2:1 37N 22 121W 47 5.6
1955 10 24 4:10 37N 58 122W 3 5.4
1957 3 22 19:44 37N 40 122W 29 5.3
1959 3 2 23:27 36N 59 121W 40 5.2
1960 1 20 3:25 36N 47 121W 31 5.1
1961 4 9 7:23 36N 43 121W 18 5.3
1961 4 9 7:25 36N 45 121W 22 5.2
1962 6 6 17:50 39N 4 123W 20 5.2
1964 11 16 2:46 37N 2 121W 45 5.1
1967 12 18 17:24 37N 3 121W 45 5.2
1969 10 2 4:56 38N 28 122W 41 5.5
1969 10 2 6:19 38N 28 122W 41 5.4
1974 11 28 23:1 36N 55 121W 30 5.1
1977 11 22 21:15 39N 26 123W 15 5.0
1979 8 6 17:5 37N 5 121W 28 5.9
1980 1 24 19:0 37N 50 121W 48 5.8
1980 1 27 2:33 37N 51 121W 47 5.5
preferentially favored over single-station
magnitudes, especially during the period from
1936-1973. This has occasionally resulted in a
change in magnitude value of « unit relative to
those values listed in Bolt and Miller (1975)
and Real et al. (1978).
As the uniformity of the listed magnitudes
with time is essential for some aspects of the
analysis that follows, it is critical that the
ioo
IO
M•.•,O
ø'ø'1
• 19.07 1910 I .cj•O• .F•10 • YEAR
0,001 t , , , i , i • •. i
0.1 1.0 I0 I00 I000 I0,0•
DAYS AFTER 1906 EARTHQUAKE
FIG. A1. Frequency of earthquakes felt in
Berkeley, California. Data from 1906-1907 from
Lawson et al. (1908). Data from 1910-1949 from
Bolt and Miller (1975). Reference line is pro-
portional to t -1. Earliest time interval (arrow)
extends to time of 1906 mainshock. Mean pro-
duction rate of M>3 and M>4 earthquakes occurring
within 50 km of Berkeley area also shown.
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
intensity magnitudes (M I) are not greatly
biased relative to the instrumental magnitudes
(ML). A direct comparison of M I with M L for 19
events that occurred between 1945 and 1969
shows that M I- ML= -0.2 + 0.2. Thus, there is
a suggestion that magnitude values based
exclusively on intensities are marginally
smaller than M L. There is also some evidence
that M I values determined from the total felt
area are systematically larger than those deter-
mined from the M. M. intensity V area (0.25 _+
0.25). As all computed intensity magnitudes
were averaged by Toppozada et al. (1979) in
preparing their catalog, we suspect that magni-
tudes from the pre-instrumental period may be
underestimated by about 0.3. However, these
magnitudes would have to be overestimated by
1.0 magnitude units to affect our conclusions.
This appears to be highly unlikely.
The spatial uniformity of the earthquake
catalog is of similar concern to us. We
believe that it is reasonable to assume that
the catalog attains its threshold reporting
level about the time when local newspapers were
established. This would correspond to about
1855 for the region from San Jose to Santa
Rosa, 1860 for the region from Monterey to
Point Arena, and about 1900 for the region
north to Cape Mendocino. Magnitude 5 earth-
quakes are present in the catalog by 1865 from
south of Monterey to Santa Rosa, and as far
north as the latitude of Point Arena by 1870.
Earthquakes located at the western edges of the
San Joaquin Valley are also present in the
catalog by the early 1860's.
Acknowledgements. We thank Mike Lisowski for
the use of his unpublished analysis of the
Hayward net. The critical comments of Tom
Hanks, Wayne Thatcher and Dave Boore were of
great value to us in preparing this paper.
Graphics by C.R. McMasters.
References
Addicot, W.A., Late Pliocene mollusks from
San Francisco Peninsula, California, and
their paleographic significance, Proc.
California Acad. Sci• Fourth Ser, 37, 57-93,
1969.
Aki, K., Possibilities of seismology in the
1980's, Bull. Seismol. Soc. Amer., 70, 1969-
1976, 1980.
Allen. C.R., The modern San Andreas fault, in
The Geotectonic Development of California,
Ernst, W.G., ed., Prentice-Hall Inc., New
Jersey, 511-534, 1981.
Bolt, B.A., The focus of the 1906 California
Earthquake, Bull Seismol. Soc. Amer., 58,
457-471, 1968.
Bolt, B.A., and Miller, R.D., Catalogue of
earthquakes in northern California and
adjoining areas 1 January 1910 - 31 December
1972, Seismographic Stations Univ. Calif.,
Berkeley, California, 567 pp, 1975.
Bolt, B.A., Lomnitz, C., and McEvilly, T.V.,
Seismological evidence on the tectonics of
central and northern California and the
Mendocino escarpment, Bull Seismol. Soc.
Amer., 58, 1725-1767, 1968.
Boore, D.M., Strong-motion recording of the
California earthquake of April 18, 1906, Bull
Seismol. Soc. Amer., 67, 561-579, 1977.
Cummings, J.C., The Santa Clara formation and
possible post-Pliocene slip on the San Andreas
fault in central California, in Dickinson, W.
R., and Grantz, A., eds., Proc. Conf. Geol.
Prob. San Andreas Fault S¾stem• Stanford Univ.
Publ. Geol. Sci., 11, 191-207,1968.
Ellsworth, W.L., and Marks, S.M., Seismicity of
the Livermore Valley, California region 1969-
1979, U.S. Geol. Surv. Open-File Report
80-515, 41 pp, 1980.
Fedotov, S.A., Regularites in the distribution
of strong earthquakes in Kamchatka, the Kuril
Islands and northeastern Japan, Akad. Nauk.
SSSR Inst. Fiziki Zeml:Trudy, 36, 66-93, 1965.
Fedotov, S.A., The seismic cycle, quantiative
seismic zoning, and long-term seismic fore-
casting, in Seismic Zonin• of the USSR• S.
Medvedev, ed., Izdatel'stvo "Nauka", Moscow,
1968.
Gutenberg, B., and Richter, C.F., Seismicity of
the Earth and Associated Phenomena, Hafner
Publ. Co., New York, 310 pp, 1954.
Hanks, T.C., and Kanamori, H., A moment magni-
tude scale. J. Ge,ophys. Res, 2348-2350, 1978.
Herd, D.G., Neotectonic framework of central
coastal California and its implications to
microzonation of the San Francisco Bay Region,
i•n Brabb, E. E., ed., Progress on Seismic
Zonation in the San Francisco Bay Re•ion, U.
S. Geol. Surv. Circ. 807, 3-12, 1979.
Imamura, A., Preliminary note on the great
earthquake of southeastern Japan on September
1, 1923, Bull. Seismol. Soc. Amer., 14,
136-149, 1924.
Kelleher, J., and Savino J., Distribution of
seismicity before large strike slip and
thrust-type earthquakes, J. Geophys. Res.,
80, 260-271, 1975.
Lachenbruch, A.H., and Sass, J.H., Heat flow
and energetics of the San Andreas fault zone,
J. Geophys. Res., 85, 6185-6222, 1980.
Lawson, A.C., Chairman, California EarthQuake
of April 18• 1906• Report of the State Earth-
Quake Investisation Commission, Vol. I.,
Carnegie Institution of Washington, 451 pp.,
1908.'
Lehner, F.K., and Li, V.C.F., On the stressing
of the lithosphere in an earthquake cycle
(abs), EOS, Trans. Amer. Geophys. Union, 61,
1051, 1980.
Louderback, G.D., Central California earthquakes
of the 1830's, Bull. Seisim. Soc. Amer, 37,
33-74, 1947.
Mavko, G.M., Simulation of creep events and
ELLSWORTH ET AL. 139
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
earthquakes on a spatially variable model
(abs), EOS• Trans. Amer. Geophys. Union, 61,
1120-1121, 1980.
McCann, W.R., Nishenko, S.P., Sykes, L.R.,
and Krause, J., Seismic gaps and plate
tectonics: seismic potential for major plate
boundaries, U.S. Geol. Surv. Open-File Report
78-943, 441-584, 1978.
Minster, J.B., and Jordan, T.H., Present day
plate motions, J. Geophys. Res., 83,
5331-5354, 1978'•
Mogi, K., Study of elastic shocks caused by the
fracture of heterogenous materials and its
relations to earthquake phenomena, Bull.
Earthq. Res. Inst., 40, 125-173, 1962.
Mogi, K., Some features of recent seismic
activity in and near Japan (1), Bull. Earthq.
Res. Inst., 46, 1225-1236, 1968.
Olson, J.A., Lindh, A.G., and Ellsworth W.L.,
Seismicity and crustal structure of the Santa
Cruz Mountains, California (abs), EOS• Trans.
Amer. Geoph¾s. Union, 61, 1042, 1980
Orozco y Berra, D.J., Seismologia: efemerides
seismicas Mexicanas, Memorias de la Sociedad
Cientifica, •, 258, 1887.
Prescott, W.H., The accommodation of relative
motion along the San Andreas fault system in
California, Ph.D. Thesis, Stanford Univ.,
195pp, 1980.
Prescott, W.H., Lisowski, M., and Savage, J.C.,
Geodetic measurement of crustal deformation on
the San Andreas, Hayward and Calaveras faults
near San Francisco, California (abs), EOS•
Trans. Amer. Geophys. Union, 61, 1042, 1980.
Real, C.R., Toppozada, T.R., and Parke, D.L.,
Earthquake Catalog of California January 1,
1900 - December 31, 1974, Calif. Div. Mines
and Geol. Special Pub 52, 15 pp, 1978.
Reid, H.F., The Mechanics of the Earthquake,
Vol. II of The California Earthquake of April
18• 1906, Carnegie Institution of Washington,
192 pp, 1910.
Savage, J.C., A possible bias in the California
state geodimeter data, J. Geophys. Res., 80,
4078-4088, 1975.
Savage, J.C., Comments on "Strain accummulation
and release mechanism of the 1906 San
Francisco earthquake by Wayne Thatcher, J.
Geoph¾s. Res., 83, 5487-5489, 1978.
Savage, J.C., and Burford, R.O., Discussion
of paper by C. H. Scholz and T. J. Fitch,
'Strain accumulation along the San Andreas
fault', Jour. Geophys. Res., 78, 832-845,1971.
Savage, W.U., and McNally, K.C., Moderate
earthquake seismicity in central California
1936-1973 (abs), Earthquake Notes, 45, 33, 1974.
Scholz, C.H., The frequency-magnitude relation
of microfracturing in rocks and its relation
to earthquakes, Bull. Seism. Soc. Amer., 58,
399-415, 1968.
Shimazaki, K., and Nakata, T., Time predictabel
recurrence model for large earthquakes,
Geophys. Res. Lett., •, 279-282, 1980.
Templeton, E.C., The central California earth-
quakes of July 1, 1911., Bull. Seismol. $oc.
Amer., •, 167-169, 1911.
Thatcher, W., Strain accumulation and release
mechanism of the 1906 San Francisco earth-
quake, J. Geophys. Res., 80, 4862-4872, 1975a.
Thatcher, W., Strain accumulation on the
northern San Anderas fault zone since 1906,
J. Geophys. Res., 80, 4873-4880, 1975b.
Thatcher, W., Reply to comments by J.C. Savage,
J. Geophys. Res., 83, 5490-5492, 1978.
Thatcher, W., Systematic inversion of geodetic
data in central California, J. Geophys. Res.,
84, 2283-2295, 1979.
Tocher, D., Seismic history of the San Francisco
region, Calif. Div. Mines Special Rep 57,
39-48, 1959.
Toppozada, T.R., Earthquake magnitudes as a
function of intensity data in California and
western Nevada, Bull. Seism. Soc. Amer., 65,
1223-1238, 1975.
Toppozada, T.R., Real, C.R., Bezore, S.P., and
Parke, D.L., Compilation of pre-1900
California earthquake history, Calif. Div.
Mines Geol., Open File Release 79-6, 271 pp,
1979.
Toppozada, T.R., Real, C.R., Bezore, S.P., and
Parke, D.L., Preparation of isoseismal maps
and summaries of reported effects for pre-1900
California earthquakes, Calif. Div. Mines
Geol. Open File Release 80-15, 78pp, 1980.
Townley, S.D., and Allen, M.W., Descriptive
Catalog of earthquakes of the United States
1769 to 1928, I. Earthquakes in California
1769-1928.•Bu11. Seism. Soc. Amer., 29,
21-252, 1938.
Turcotte, D.L., Stress accumulation and release
on The San Andreas fault, Pageoph, 115,
413-427, 1977.
Weeks, J., Lockner, D., and Byerlyee, J., Change
in b-value during movement on cut surfaces in
granite, Bull. Seism. Soc. Amer., 68, 333-341,
1978.
Webber, G.E., and Cotton, W.R., Geologic
investigation of recurrence intervals and
recency of faulting along the San Gregorio
fault zone, Final Tech. Rep., U.S.G.S. Grant
14-08-0001-16822, 135 pp, 1980.
Webber, G.E., and Lajoie, K.R., Late Pleis-
tocene and Holocene tectonics of the San
Gregorio fault zone between Moss Beach and
Point Ano Nuevo, San Mateo County, California
(abs), Geol. Soc. Amer. Abst. with Prog., •,
524, 1977.
Willis, B., Earthquake risk in California 8.
earthquake districts, Bull. Seismol. Soc.
Amer., 14, 9-25, 1924.
Wood, H.O., On the region of origin of the
central California earthquakes of July,
August and September, 1911, Bull. Seism. Soc.
Amer., 2, 31-39, 1912.
140 ELLSWORTH ET AL.
Maurice Ewing Series
Earthquake Prediction: An International Review
Vol. 4
Copyright American Geophysical Union
