Several years ago I wrote a requested review paper on the effects of electrical and
geomagnetic effects on the occurrence of earthquakes for a conference. The
paper was never published, but may be of some interest to readers of this board
as it should be at a level which most can understand. I am attaching it in several parts below. It contains observations on mechanisms, research and
earthquake lights. Perhaps it will help to facilitate a more informed conversation
on these subjects on this board.

REVIEW PAPER ON ELECTOMAGNETIC MECHANISMS AND EARTHQUAKE PHENOMENA

By: Dr. Lowell Whiteside

INTRODUCTION

The major force preventing faults from moving in response to
tectonic forces is friction. Fault friction is a function of normal
stresses acting on the fault which vary in response to changes in
local time-variable conditions such as temperature, pore pressure,
loading and unloading, time of contact between the sides of the
fault, and electrical charges on the fault. Each of these
parameters may also vary as the state of stress on the fault varies
or as external environmental conditions vary. May of these
properties are reversible, i.e. while strain on a piezoelectric
mineral may cause a current to flow, a flowing current will also
change the stress field of the same piezoelectric mineral. An
association between transient behavior of rocks on or near faults
and earthquakes may be fault-precursory (due to changes in the
rocks), triggering (due to external forces acting on the fault), or
a combination of these. a major problem exists, therefore, in using
dynamic parameters as precursory phenomena, since all factors which
are not directly related to the condition of the fault must be
eliminated.
Some of the parameters which have been suggested as changing
fault friction in time scales useful in short-term earthquake
prediction are: gravimetric loading or unloading due to reservoir
filling (Kisslinger, 1976; Nikolayev et al., 1976); temperature
variations (Popov, 1966; Yoshioka, 1985; Ismail-Zade et al., 1982);
polarization due to S-wave triggering (Kisslinger, 1976; Wesson and
Nicholson, 1988); Lunar and Solar Tides (Kilston and Knopoff, 1983,
Sadeh and Wood, 1978, Ulbrich, Ahorner and Ebel, 1987, Zhong-Yi et
al., 1983, Shirley, 1988 and many others), traveling stresses
(Guberman, 1981; Guberman and Rotvayn, 1986; Johnston, 1989),
changes in pore pressure and fluid content of rocks caused by
changes in the stress of the rocks (Hollister and Weaver, 1968,
Crampin, 1987); seismoelectromagnetic effects (e.g. Physics of the
Earth and Planetary Interiors, Special issue, Oct. 1989); and fault
loading due to atmospheric pressure variations (e.g. Yamauchk,
1987; Muller and Zurn, 1983; Namias, 1988, Mil'kis, 1986). Here I
consider only aspects of short-term seismoelectromagnetic effects.
I will attempt to show, using teleseismic data, that changes
in the external electromagnetic field in the fault region are
statistically correlated with higher seismic activity and in
particular on the Aleutian subduction zone and in Central America.
Changes in electrical parameters can reduce the strength and
viscosity of the rocks and lower normal and shear stress on the
fault, thus reducing friction. In addition, I will show that much
long- and short-period seismic variation is closely correlated with
electromagnetic variations in the Aleutians, portions of Central
America and in aftershock sequences in California. The reaction of
the fault to external environmental changes therefore must be
considered before confidence can be gain in observed fault
precursory activity.

PREVIOUS WORK

Laboratory Experiments:

Extensive studies of the electromagnetic properties of rocks
undergoing stress have been conducted. These analyses have examined
a wide range of electrical and magnetic properties, primarily of
concern as precursory signals to earthquakes.
Yamada, Masuda and Mizutani (1989) investigated anomalous
electromagnetic (EM) emissions from rocks. The considered the three
possible explanations offered by Ogawa et al. (1985) for the
observed emission of EM radiations with it's main power in
frequency ranges > 0.5 MHz. 1) electrification on contact or
separation, including triboelectrification; 2) streaming potential
electrification (electrokinetic effect) involved in microcracked
rocks containing capillary water and 3) piezoelectricity. They
observed that striking a rock with a pointed bar produces E<
emission with amplitude directly related to the size of the surface
cracks produced. They combined this observation with the fact the
EM emission has been observed from non-quartz bearing rocks,
although amplitudes of this emission are less than from quartz-
bearing rocks, to conclude that piezoelectricity is not the cause
of the emission. Likewise, since they used "dry" rocks, they
dismissed streaming potential as a cause for EM emission. Goupheld
et al. (1987) argued that elastic wave propagation could excite EM
emissions. Yamada, Masuda and Mizutani, however, observed
simultaneous origins for both elastic (Acoustic Emission (AE)) and
EM radiation. Since the acoustic waves are thought to be coming
from the rapid growth of cracks. they concluded that the EM
emissions were generated at the time of microcrack generation.
These conclusions, however must be questioned, since crack growth
coincides with stressing of the minerals surrounding the crack and
these minerals may respond with piezoelectric activity. The
implication that quartz is the only piezoelectric mineral in rock
is also incorrect. Any mineral which is asymmetric about an axis of
symmetry will show piezoelectric behavior. Rock salt, hornblende
and many other common rock forming minerals also show this effect.
Experiments by Shevcova (1984) and Khatiashvili and Perel'man
(1989) confirm that much of the electromagnetic radiation emitted
prior to fracturing in laboratory experiments is caused by the
breakage of ionic bonds during crack formation. They, however, also
report that EM emissions of this type may be detected up to 1000 km
from the site of the future earthquakes and do not necessarily
maximize at the epicenter. In fact, these researchers agree with
Yamada, Masuda and Miztani (1989), that fresh crack formation is
required for these emissions, and that this is not likely on the
main fault prior to a large mainshock. The technique developed in
1989 by Takahashi and Takahashi for predicting imminent earthquakes
from emitted EM by tomography may be of limited usefulness since
the timing and locations os this phenomena do not necessarily
correspond with and eventual rupture.
Balbachan and Tomashevskaya (1987) have conducted experiments
on rock strength upon the application of an electric field. These
experiments show that in laboratory specimens of marble, basalt,
quartzite and diabase, rock strength is reduced by up to 50%
through the induction of an electric charge by friction on dry
rocks and of an order of magnitude higher when water is present.
This reduction in strength relaxes over 3 weeks time, suggesting it
is caused by generation of a charge mosaic between the rock grains,
and not by mechanical crack formation from the frictional rubbing.
In earlier experiments Balbachan and Parkhomenko (1983) had found
that the relaxation of surface charge occurs in sudden bursts with
maxima in limestone at 25 to 160 hours and a minima between 60 and
80 hours after the initial rupture. Some changes in seismicity of
aftershock sequences in California occur near these times (Creamer,
1987) and may be related to these observed bursts of discharging.
In general, from the work of Balbachan and Tomashevskay (1987) it
appears that the electrification of the specimen promotes the
development of microfractures under loading, or redistributes
charged lattice defects, leading to a lessening of strength in the
rock.
Zaborovskiy et al. (1968) have investigated the properties of
rocks in electromagnetic fields whose frequencies vary from 104 to
106 Hz. They find that application of a strong electromagnetic
field has several effects on rock-forming mineral.s At particular
radio frequencies polarization of rock molecules to ionic form
occurs and electrical conductivity increases tenfold as electrons
are released from the matrix. The effect of increased ionization on
stressed rock where cracking is occurring is to accelerate the
crack-forming process which depends on bond breaking at crack tips.
The result of this is a tendency for rocks not only to emit radio
frequencies just prior to fracture but also to increase fracturing
in response to radio signals. Under these conditions cascading of
cracks is possible in pre-stressed rocks subject to strong radio
signals in which molecules are excited to polarization. This effect
is especially prominent when water is present.
Maeda (1983) has observed triboluminescence (TL) during
frictional sliding of rock samples in the laboratory. TL is shown
to increase exponentially with normal stress and with increased
surface area of contact and completely disappears when the external
force is removed, indicating plain friction is insufficient to
maintain TL. Some rocks, however maintain TL for up to 10 seconds
after sliding. This may be an interrelation between charge on
individual grains and inhomogeneous temperature variation
(Yoshioka, 1985).

OBSERVATIONS

Electromagnetic Emissions:

Laboratory experiments (see above and Cress et al. 1987)
clearly demonstrate that EM emission should be expected as part of
the preparation process of large earthquakes. The processes
preparatory to seismic events can be a source of quasistationary
currents whose density is estimated at up to 10 A/m2. Because of
the inhomogeneous nature of the earth's conductivity, however (for
example, conductivity in the material filling faults can be several
times greater than the surrounding rocks) locally, these electric
fields can be much high in intensity.
Electromagnetic emissions (EME) have been observed by Yoshino
and Tomizawa (1989) in connection with a volcanic eruption of
November, 1986 at Mt. Mihara in Japan. Emissions in the 81-82 kHZ
range were theorized to be caused by magmatic intrusion causing
cracking, which in turn leads to EME as described by Cress et al.,
(1987) from laboratory experiments. Observations from Gokhberg et
al. (1982) find emissions also in the 81-82 kHz range occurring
prior to magnitude 5.5 - 6.5 earthquakes on 25 September, 1980 and
28, January, 1981 in Japan.
EME from earthquakes have been continuously observed by
Nikiforova et al. (1989) since 1978 in the Issyk-Kul seismic region
of Kirgistan. They report EME with frequency of 15 kHz occurring 2
days to 6 hours prior to earthquakes in the Kirghistan region. They
also observe anomalous EME for larger earthquakes at distances >
400 km at their field sites. The following features are reported
as occurring with EME in Khirgistan:

1) regular daily behavior of EME breaks down.
2) Intensity of regular maxima exceeds normal by more than two
standard deviations.
3) Greatest disturbances in EME also precede earthquake shocks
4) EME disturbances last from a few hours to several days.,
5) EME is normally only associated with deep earthquakes.

Oike and Ogawa (1986) operated a similar continuous
observation of EME in Japan using a radio as a sensor from 1983 to
1985. They found for inland earthquakes a close association between
anomalous spikes in EME in the LF range (163 kHz) and shallow
earthquakes. The length of the period of LF anomaly appears
directly proportional to the magnitude of the mainshock in Japan,
suggesting the observed rations are closely related to the
expansion of fractures in the focal region. We will see that other
explanations of Nikiforova et al.s; data area possible.
Strong precursory EME occurred several hours prior to the 1989
Loma Prieta earthquake and in association with the 1993 Guam
earthquake of 8 August (Mw 8.1).
One of the first observation of EME associated with
earthquakes were extraordinary radio emissions that were observed
by radio observatories at Boulder Co., Lake Angelus, Michigan,
Sacramento Peak, New Mexico and at Makapuu Point, Hawaii on May 14-
15 1960 about six days before the great Chilean earthquake of May
21, 1960 (Mw=9.6). The signals were tuned to 18 Mhz, were diffuse
and were interpreted as coming from the Chilean earthquake by
Warwick et al. (1982). It should be noted, however, that one of the
largest solar proton storms encountered in the past 50 years began
on May 13 at 8:30 with particle energies >> Mev and lasted through
May 15 (Svestka and Simon, 1975). Proton storms of this magnitude
can create very anomalous radio signals.
Many reports of FM radio noise preceding earthquakes have also
been reported. To cite a few examples, Markert (1976) observed VHF
noise along the whole FM radio band several minutes to seconds
before a severe quake in northeast Italy in 1976. Derr (1973)
reports that Sr. Octavio R. Garcia A. of Mexico City reported a
strong noise on all FM stations of his car radio after the July 16,
1973 earthquake in Guerrero, Mexico (Ms=5.7). In October, 1998, a
radio station operator, David Johnson in Peoria Illinois called the
National Geophysical Data Center to report a strong anomalous
signal in the 15-20 kHz range occurring at the time of the call.
The call was received one minute following the calculated
hypocentral origin time of a Mw 7.2 earthquake in Peru (David
Johnson, personal communication).
Investigation of the human reaction to certain VLF
electromagnetic radiation indicate that in some people and many
animals, the ear is directly stimulated by VLF radiation without
the intermediary of acoustic waves. The mind interprets this
interaction as sound (Condon, 1969, Hughes, 1975). In addition to
stimulating the sensing hairs of the ear, VLF fields can also
excite surface acoustic waves in environmental objects which can be
heard by some (Keay, 1980). The sound produced by direct
stimulation is directionless, often appearing to emanate from a
position about 30 cm. above the head and is heard as a high pitched
hum, buzz or clicking. Auroras are commonly "heard" in this manner
(Silverman and Tuan, 1973), as are meteors (Keay, 1980). Direct
perception of radio waves could explain some anomalous animal
behavior, which is similar to that observed before electrical
storms as well as hissing or buzzing noises some people "hear"
before earthquakes. An alternative explanation for these noises may
be infrasound, very long wavelength sound which can be heard by few
and often can be sensed only with elaborate equipment (Greene,
1973). Some animals are particularly sensitive to infrasound
however, elephants, for example, use infrasound to communicate over
separation distances of several kilometers.

Luminous Activity Associated with Earthquakes:

Earthquake lights (EQL) are perhaps the most commonly cited
manifestation of electrical phenomena preceding or coseismic with
an earthquake. An early review of the observations and theories
regarding these phenomena was provided by Derr (1973). EQL are
described as either radiating from a ground-based source or as a
general glow of the sky, sometimes appearing as a luminous body 30
to 200 m. in diameter. EQL's usually are coseismic or follow
earthquakes, lasting up to two minutes, although some have been
reported occurring prior to earthquakes, and some several days
after earthquakes. EQL are often associated with the passage of
low-pressure systems, and it is this fact that leads Grigoryev et
al. (1989) to suggest that some earthquake lights may be caused by
EME exciting water molecules in the air to instability, and
releasing photons of visible light in the process. This would
explain many of the reports of EQL at sea (e.g. Byerly, 1942) which
are impossible to explain using triboluminescence of piezolelectric
causes. If water molecules are polarized by EME in this manner the
observation of sudden storms and winds associated with earthquakes
could be explained (Stewart, 1902). Strange fogs and mists such as
that associated with the 4 February, 1975 (Mw=7.5) which covered
the soil from 2 to 3 meters, followed exactly the strike of the
fault that created the shock, was layered, and a strong odor, and
appeared 1-2 hours prior to the earthquake (Corliss, 1983) might be
explained.
Current theories being considered for the generation of EQL
begin with the possibility that large-scale electrostatic fields
emerge in the preparation zone. Observations cited by Tzerfas
(1971) and Derr (1973) such as near-surface glow of air before
earthquakes, breakdown of electric cables and spontaneous switching
on of luminescent lights or those of Sadovsky et al. (1981) in
which he reports that the March 4, 1977 Romanian earthquake was
accompanied by numerous electrical problems in computers, telephone
and telegraphic equipment along with EQL suggest that occasionally
such intense electric fields may occur. Derr and Persinger (1986)
suggest that nocturnal lights prior to earthquakes are caused by
peizoelectric effects from small fractures, but at the time of
rupture, EQL are caused by triboluminescence as the strain on the
fault is released. Brady and Rowell (1986), however show that while
light is emitted from gas surrounding stress-fractured rock, the
spectrum differs from that observed in EQL. Warwick et al. (1982)
shows that the field calculated for the shearing of California
granite (3 X106 V/m) is slightly less than that required for
breakdown of air molecules, suggesting that at least for this rock
type, other causes of airglow must be pursued. It is of interest
that all of the classical EQL events noted by Derr (1973) were
preceded within two days by major solar activity and disturbed
geomagnetic field conditions.

Other Electric and Magnetic Effects Associated with Earthquakes:

Some electromagnetic activity associated with earthquakes
occurs in the ionosphere. Fishkova et al., (1985) reports that the
intensity of the green oxygen emission, which characterizes the
dynamics of the ionospheric "E' layer increases several hours prior
to earthquakes on the nearby Trans-Caucasus fault. Nestorov (1979)
observed unusual fading of FM radio traces 1 hour before the
Vrancha earthquake. Gokhberg (1987) reports similar findings from
the 'Omega' navigation system which last about one hour and precede
earthquakes occurring below the satellite's path. Sobolev
Husamiddinov (1985) report a 20% increase in the night-time
ionosphere FM emissions at times when nearby seismicity increases.
If these listed effects are caused by earthquake preparation
in the fault zone, the atmosphere and ionosphere must be extremely
sensitive to the small electromagnetic changes which are
theoretically possible in the preparation zone. Observations of
the ionosphere from GOES-2 and reported by Parrot and Mogilevsky
(1989) suggest that conditions before earthquakes near Djibouti
(11N, 42E), may cause the observed dramatic increase in electron
density in the Es layer before the shock and increased electric and
magnetic potential after the earthquake in the ionosphere in two
ways: 1) by an increase in d.c. electric and electromagnetic field
due to cracking in the earthquake preparation process; and by
emission of VLF electromagnetic waves which propagate upward into
the magnetophere. Gokhberg et al. (1984) and Alimov et al. (1989)
showed mechanisms by which an electric field produced in a large
area around an earthquake epicenter gives rise to a density
increase in the Es layer and to electromagnetic wave in the
ionosphere. After the earthquake, seismic waves may generate
pressure waves which propagate upward in the atmosphere and become
amplified with height because of the rarefaction of the atmosphere
with height.
In addition to the effects of earthquake preparation and
rupture on electric fields, magnetic field changes have also been
noted, largely in the earthquake preparation zone. These
perturbations of the magnetic field are probably related to micro-
crack formation and the resulting electrification of grain
boundaries. Models by Gershenzon et al. (1989) show that while
most of these effects would be localized, they may, under
particular condition of cracking and local geology, be observed
hundreds of km from the fault zone.
Mueller and Johsnston (1989) report that the acoustic wave of
the Mt. St. Helen's eruption of 18 May, 1980, was accompanied by a
traveling magnetic field disturbance measured by magnetometers
along the San Andreas fault, and caused field disturbances at
distances up to 2000 km. A similar magnetic field disturbance was
observed during the eruption of Mont Pelee on 8 May, 1902. In this
case magnetic needles were disturbed at magnetic observatories from
the Hawaiian Islands to England (more than 5000 km from Martinique)
within 5 minutes of the eruption (Baure, 1902). Johnston (1989)
suggests that such traveling ionospheric waves may explain many of
the unusual ionospheric perturbations apparently related to
earthquakes. Harmonic changes in the horizontal and vertical
components of the magnetic field changes observed at Port Moresby,
Australia after the 21 May, 1960, Chile earthquake were observed to
correspond with the three most important torsional free-earth
oscillation modes at 43.6, 28.2 and 21.6 minutes (Winch et al.,
1963) and following the 1964 Alaska earthquake (Mw 9.5) with
similar periods to OT8 at Lebanon New Jersey for several days after
the earthquake (Hirshberg and Currie, 1968).
One hour and 6 minutes before the Alaska earthquake, the
magnetic field was observed to suddenly increase by 100 gamma
(Moore, 1964). Moore suggested this was a piezomagnetic effect in
the nearby rocks undergoing stress leading to speculation that
changes in local magnetic field could be used as precursors to
earthquakes. Disappointingly, after ten years of monitoring of the
magnetic field in California, Johnston (1989) concluded that
coseismic transient magnetic and telluric field gradients as below
instrumental precision, and that the absence of significant changes
in telemetered magnetometers and near-fault telluric currents
suggest magnetic and electric fields do not routinely occur near
active faults and cannot be used reliable as earthquake precursors
in California.
Rikitake (1987) summarizes the magnetic and electric signals
precursory to Japanese earthquakes. He identified five cases where
geomagnetic field variations were observed prior to earthquakes,
ten with earth current precursors, 39 where anomalies were
recognized in the resistivity of the fault zone prior to mainshock
and seven where EME was a recognized precursor. In all these cases,
he finds a positive correlation between the amplitude of the
precursor and the subsequent magnitude of the associated
earthquake. On the basis of his analysis Rikitake concludes that
each of these precursory signs is indeed a measure of crustal
strain.

Earthquake Prediction Attempts:

A number of attempts to use electromagnetic phenomena in
short-term earthquake prediction schemes have been made. The
notorious work of Varotsos and Elexopoulos (1984a, 1984) is an
example confined to electrotelluric currents in the earth. This
team has monitored the telluric sites in Greece since 1981 and
searched for precursory seismic electric signals (SES). Their main
conclusion is: "every sizeable earthquake is preceded by an SES and
inversely every SES is always followed by an earthquake, the
magnitude and epicenter of which can be reliably predicted." Such
signals are observed 6-115 hours before quakes. While these
researchers claim .99 significance to their results (Meyer et al.
1985), there has been considerable discussion of their results and
many researchers disagree with their statistical conclusions.
Chinese researchers have found that reversals in magnetic
field polarity at geomagnetic observatories appear to occur within
one month of succeeding large earthquakes (XXX). These changes can
be mapped over geographical fronts. Many of these reversals are
directly associated with large solar-generated geomagnetic storms.
Most attempts to predict earthquakes on the basis of electric
or magnetic effects have had limited success, for even if reliable
precursors exist which relate to the seismic preparation process,
they must be differentiated on the recording instruments from
electric and magnetic effects caused by cyclical variations in the
earth's stationary magnetic field, transient variations in the
geomagnetic field caused by charged particles in the solar wind,
proton storms, both of which also enhance auroras and sudden
ionospheric disturbances (SID) which induce currents on the earth
and produce radio waves and interference. These external causes of
electromagnetic disturbance have all been suggested as triggers for
earthquakes -so how can the trigger be differentiated from the
precursor.

Solar-Terrestrial Electromagnetic Effects:

The concept that seismicity may be influenced by solar
activity has not generally been received favorably by
seismologists. This may be due to the many attempts to correlate
solar cycles with everything from the price of hogs to Napoleonic
wars. If the sun influences seismic activity it does so only as a
short-term moderation of the far more powerful plate tectonic
forces.
Simpson (1967) showed that on a day-to-day basis, the number
of earthquakes in the NEIS catalog reflected the number of solar
flares. His hypothesis is that magnetic coupling between the solar
plasma and the geomagnetic field may impose torques that alter the
rotation velocity of the earth and hence induce faulting on weak
zones that are already under stress. Other observational data
supports this hypothesis. An abrupt slowing of the earth's daily
rotation of 0.79 seconds corresponded with a great flare of 20
July, 1959 (Schatzman, 1960 and Danja, 1962). Danja showed that
during times when solar activity as quiet, rotation remained steady
or decayed gradually, but during active solar periods, the rotation
slowed in jerks which corresponded with the time of major flares.
If Simpson's hypothesis is correct, torsional free-earth
oscillations would be expected to be initiated independently of
great earthquakes at the time of major flares.
The earth's magnetic field responds gradually to solar flares
with the maximum influence occurring with a 3 to 4 day lag time in
86% of the cases (Bell, 1964). This is generally due to the travel
time of the solar wind following the flare. The solar particle flux
creates geomagnetic storms in the ionosphere as part of it's
interaction with the earth environment. Simpson (1967) also noted
that there was a lag time of 3-4 days between flare activity and
increased seismic activity.
The efficiency of the effect of flares on the earth is not
continuous. Satellite observations have shown that the
interplanetary magnetic field is separate into oppositely polarized
adjacent quadrants. Within each sector, the polarity of the
interplanetary field is predominantly in one direction. The
interplanetary field lines are rooted in the sun and hence rotate
with an approximate 27-day period, although this period various
depending on planetary conjunctions and oppositions and the tidal
effects of planets. The sector structure is extended outward by the
solar wind. The boundaries are very thin, approaching the atomic
radius in thickness and hence times at which such boundaries sweep
past the earth are well defined (Wilcox, 1975). Besides the sharp
change of polarity at the boundary, there is also a large-scale
pattern. During the several days before a boundary is crossed by
the earth, magnetic conditions on the sun, in interplanetary space
and in the terrestrial environment are quieter than usual, this
minimum is reached about a day before the boundary crossing,
activity then increases suddenly and reaches a peak within the next
two days. At the same time, the Van Allen belts and the ionosphere
in general "breathe" inward and outward as the sector structure
passes the earth. Sector boundaries can be anticipated four to five
days in advance, since that is the transit time of solar wind
plasma from the sun to the earth. A list of solar sector boundary
crossings is available from 1957 to 1975 (Shapley et al, 1975).
Wilcox (1975) showed that the boundary crossing was reflected in
weather patterns, an analysis which was verified by Hines and
Halevy (1977).
There are several aspects of solar flares which can
theoretically affect the terrestrial system. First, solar flares
are usually associated with intense radio signals with frequencies
ranging from 10's of hertz to VHF bands. These radio bursts can be
extremely powerful, reaching flux values of more than 10-16 W/m2Hz.
The work of Zaborovskiy et al. (!968) shows that such EM could have
the effect of expanding cracks and even triggering earthquakes.
This would be an immediate effect, and observations of changes in
seismicity would be nearly simultaneous with the flare.