These results suggest that triggering of aftershocks is
occurring due to SFE. A further test of the relation between solar
flares and aftershocks in the Imperial Valley sequence can be
conducted by comparing periodicities in seismicity with known
periodicities of solar flares identified by Rieger et al. (1984)
and Ichimoto et al. (1985). To identify periodicities in
earthquakes in Imperial Valley (32.1N to 33.3N 115W to 116W), the
catalog of 10576 events was reduced to 4365 by imposing a minimum
magnitude of 2.0. Figure XX shows the interevent times for
Imperial Valley. The interevent residuals tend to peak near times
of known periodicities in solar flare activity (74, 92, 115 and 155
days). A similar analysis for the combined aftershocks of Imperial
Valley, Coalinga and Borrego Mt. (21914 events) is shown in Figure
XX. Enhanced seismicity occurs with periodicities shown in Figure
XX, but also near 27-28 days, the solar rotation period. For the
first 100 days, the 27-day periodicity dominates, but after 100
days, the dominant period is 155 days. Since the earthquake
periodicities and flare periodicities are independent, the
coincidence of these periodicities suggests flare-related
triggering.
It has been suggested that it SFE is a triggering mechanism
for earthquakes, and the intermediary is the geomagnetic field,
then earthquakes would be expected to occur more commonly in the
three days after solar sector boundary crossings. To test this, the
NGDC data from the Aleutians (175E to 163W for the years 1957 -
1966 was chosen to test whether these early data are sufficient to
show these effects even though Shapley et al's (1975) list of solar
crossings extends to 1975. The analysis shown in Figure XX shows
that the likelihood of an earthquake being recorded in the
Aleutians within three days of solar sector boundary crossing is
more than twice that of any other time interval. The standard
deviation on the mean is near 10, so this increase has significance
>>.999. Three hundred earthquakes (48%) occur in the Aleutians
within 3 days of solar sector crossings. During this time there
were 300 solar sector crossing,s or an average of one every 14.6
days.
The relation between solar flares and earthquakes in the
Aleutians was also investigated. Flares covering more than 3
square degrees of solar surface or of Type 2N or greater were
considered, and are considered major flares as listed in the
Quarterly Bulletin on Solar Activity of the International
Astronomical Union for 1968-1984. 408 major flares were included.
The rate of events per day difference for major flares is shown in
Figure XX, showing nearly twice as many earthquakes occurring on
the same day as flares than any other day in the 400 day period
shown. Figure XX shows a similar result for earthquakes in Central
America (107W to 79W, 1964 - 1987). Since smaller magnitude events
have teen removed from both these data sets, triggering apparently
is not occurring only in smaller earthquakes.

DISCUSSION

The results of the analysis of time series of earthquakes in
the Aleutians, Central America and other areas of the world and of
aftershock sequences in California implicate SFE as triggers for
earthquakes of all magnitudes. Several of the proposed fault-
related precursors resemble SFE. Since both these effects occur
prior to or coincident with earthquakes, confusion can occur.
An example in which SFE may be mistaken for fault-related
precursory or coseismic activity is that of earthquake lights
(EQL). Derr (1973) reports on a number of specific occasions during
which EQL were observed. However, many of these may report auroras
rather than EQL.
From the Idu Peninsula earthquake of 25 November, 1930 (JMA
M=7.0) more than 1500 reports of EQL were collected by Musya
(1931). Quoting Musya "The observations were so abundant and so
carefully made that we can no longer feel much doubt as the reality
of the phenomena (EQL) and of their connection with the shocks. In
most of them, the sky was lit up as if by sheet lightning... At one
place on the east side of Tokyo the light resembled auroral
streamers diverging from a place on the horizon... A ruddy glow was
seen in the sky." Since the description sounds mich like
descriptions of auroral displays, the interpretation that these
were EQL needs to be examined.
Data on specific solar activity is not as well documented for
years prior to the mid-1950's as for those after. The "Bulletin for
Character Figures of Solar Phenomena" (BCFSP) of the International
Astronomical Union does not begin giving hours of solar flares
until 1950. The Bulletin (1931), however, does note that on 25
November, 1930 the first H alpha solar flare of note since August
of that year occurred with a very bright eruption of importance 2.
Only 3 eruptions of this size were observed during the entire year
of 1930. The sunspot group which produced this flare was in the
central meridian of the sun and formation of a large new sunspot
group in the central meridian was occurring. These conditions were
such that the influence of the flares on the ionosphere were
maximized. The BCFSP does not give a time of initiation for this
flare, but the annual report of the Kakioka Magnetic Observatory
for 1930 (pub. 1931) shows the vertical intensity of the magnetic
field increasing from 34746 gamma to its highest value during
November 1930 of 34774 gamma between 25 Nov. at 15:00 and 25 Nov at
18:00 U.T. and disturbed magnetic conditions throughout the day of
25 November. Maximum auroral activity may easily have coincided
with the maximum magnetic disturbance. The earthquake actually
occurred at about 19:03 U.T. These observations suggest that many
if not all of the EQL associated with this event were auroras.
A second example of a possible mistaken interpretation of
simultaneous occurrence of earthquakes and aurora as EQL is
provided by the famous sequence of pictures originally from Yasui
(1968 taken by the Matashiro dentist T. Kuribayashi. This series
(at the time the only known pictures of EQL) shows a 22-second
sequence of lights associated with a small, local earthquake near
Matsushiro. Since this earthquake cannot be found in ISC or PDE
earthquake listings, it may be assumed that it occurred when the
lights were first observed (4 December, 1965, 14:48 UT). The first
sudden impulse ionospheric magnetic disturbance since 4 November
1965 was recorded at widely separated magnetic observatories
(Memambetsu, Kakioka, Kanyoa and Simosato) as commencing at 14:43
on 4 December, 1965, just five minutes before the presumed time of
the observation of the earthquake lights. Since sudden impulse
storms occur in association with aurora, these classic pictures of
EQL may simply be more auroras.
Table X compares observations of EQL (Derr, 1973) with
possible associated aurora.

Table X: Comparison of EQL events with Auroral-causing Solar
Activity

EQdate time (UT) Location Solar Activity (date/time) Type

31 Jan. 1922 13:20 No. California Jan. 29-30 Magnetic Storms

22 Oct. 1926 12:35 Central Calif. Oct 13-14 Largest Flare
from 1910-1940

25 Nov. 1930 19:03 Idu, Japan Nov 25 17-19UT Sudden Impulse

02 Nov. 1931 Hyuga, Japan Oct 31-Nov 2 Major Magnetic
storm

28 Jul. 1957 08:40 Acapulco, MEX. Jul 27 07:00 Major Proton
Jul 28 15:57 Storm

18 Aug. 1959 06:37 Hebgen Lake Aug 18 06:44 Major Flare
Aug 18 06:20 Radio burst

04 Dec. 1965 14:48 Matsushiro Dec 4 14:43 Magnetic Storm

02 Aug. 1968 14:06 Mexico City Aug 2 14:00- Major Proton
Aug 3 03:00 Storm

02 Oct. 1969 04:46 Santa Rosa Sep 30 Major Proton
California Storm

While the simultaneity of some EQL events with likely auroras
suggest that auroras can be mistaken for EQL, it certainly does not
dismiss the possibility of EQL altogether. The release of gasses,
stimulated by electrical discharges from the fault during fracture,
the possibility of VLF-emission caused polarization of air-borne
water molecules, piezoelectric or triboluminescent effects are all
likely candidates for some light emission during earthquake
rupture. It is also possible that the magnetic and electrical
disturbances which initiate aurora trigger the production of EQL
during active faulting through induced earth currents or other
mechanism. Because EQL and auroras have many of the same causes,
however, care must be taken to determine exactly what are the
causes and effects in each instance.
Relating radio emissions to precursory fault activity is also
subject to possible errors of interpretation. The evidence suggests
that radio emissions can be caused by cracking as part of the
earthquake preparation process. However, since this cracking does
not necessarily center on the fault which is about to rupture, and
no definite times lines of effects before earthquakes has yet to be
established, the use of EME as a precursor must be done with
caution. EME from faults is difficult to differentiate from
ionosphere EME, often related to SFE.

CONCLUSIONS

Short-term electromagnetic precursors of earthquakes have been
described as belonging to several classes. Preceding or coseismic
luminous activity such as earthquake lights (EQL) have been
observed during many earthquakes. This type of electrical activity
has also been observed in laboratory experiments and is commonly
believed due to 1) piezoelectric effects, 2) charging of emitted
gasses during rupture, 3) polarization of air-borne water molecules
during rupture or 4) triboluminescence. Several of the classic
observation of EQL may be aurora.
Earthquake triggering by electromagnetic effects is
theoretically possible, first from large radio bursts polarizing
molecules of water and rock in the fault region and secondly as the
results of changes in earth rotation or free-earth oscillations due
to the interaction of the geomagnetic field and fields induced at
the time of major solar and cosmic storms.
Examination of aftershock sequences in California suggest that
about one-tenth of all events in these sequences may be triggered
at the time of solar radio bursts. As many as one-third of the
aftershocks may be related to torsional effects produced on the
earth at the time of solar flares, and proton or magnetic storms.
In the Aleutians, 48% of the event of Ml>=4.0 occur within three
days of solar sector boundary crossings, and the probability of
such an earthquake occurring on the day of the boundary crossing is
double the background rate. Different area are affected differently
by SFE. Mapping of correlation of earthquakes with changes in
geomagnetic and solar parameters suggest that these correlations
are strongest near the geomagnetic equator and at high latitudes
within 3000 km of the geomagnetic poles.
In general, the fault parameters and those of earthquakes
(such as time and epicenter) are influenced by the geomagnetic
properties of the fault and transient magnetic and electrical
fields which interact with the fault on a time-variable way. Any
study of fault behavior must and of subsequent time-variable
seismicity, must therefore, not ignore the importance of
seismoelectromagnetic effects either as precursors or as triggers
to earthquake activity.