A second effect that X-ray flares have occurs as adiabatic
heating of the ionosphere, as the X-rays interact with the
molecules in this region. This heating can expand the upper
atmosphere within seconds. The troposphere has lowered by about 20
km in the past 150 years because of increasing amount of carbon
dioxide in the atmosphere. Because the ionosphere is only about 70
km above the surface, the geomagnetic effects on the earth have
likewise increased (by about 50%). This is shown by the gradual
increase in the geomagnetic AA index which has risen consistently
over the past 150 years and quiet periods are now nearly as active
as active periods were 150 years ago. If there is a triggering
effect on earthquake it is likely that it has increased long-term
over the same period. Torsional effects resulting from sudden
impulse storms would also have increased and should be observable
in earthquake catalogs.
When flares interact with the earth's geomagnetic field, they
induce ring currents around the geomagnetic equator and around the
geomagnetic poles. These electric currents induce currents in the
solid earth with peak intensities about 1000 km either side of the
geomagnetic equator and between 2000-3500 km from the geomagnetic
poles and are strongest about 3-5 days following the sudden impulse
storm. These are referred to as "solar flare effects (SFE). During
strong storms, the induced currents can be quite strong (6-10
A/km). During the geomagnetic storm of Oct 17-18 1989, the induced
currents caused a power blackout throughout eastern Canada and
melted transformers in power plants as far south as New Jersey. If
electric currents are a cause of triggered seismicity, a delay of
3-5 days should occur in increased seismicity at high latitudes and
near the geomagnetic equator.
In addition to charged particles from the sun, the ring
currents are also enhanced by strong cosmic ray storms. Because
these storms are more common during periods of low solar activity,
triggered activity is most likely during the highest and lowest
periods of solar activity.
Sytimsky (1987, 1989) has developed an earthquake prediction
technique which he claims has an excellent success rate. His
investigations indicate that strong earthquakes arise primarily
when a strong solar wind is incident on the earth, and hypothesizes
that the increased terrestrial seismicity is an effect of the
intermittent effect of solar activity on the earth's atmospheric
circulation (similar to Simpson's (1968) hypothesis). He argues
that geomagnetic disturbances precede by one to two days major
global earthquakes.
Other researchers (e.g. Jakubcova and Pick, 1987; Goldstein,
1985; Currie, 1980; Djorovic, 1985) have attempted to show
earthquake periodicities are similar to solar periodicities. These
investigators are generally concerned with long-term variations in
the solar magnetic field such as the well-known 11-year and 22-year
solar periods. Others have concentrated on the 27-day solar
rotation period, but this is extremely difficult to differentiate
from the 28-day tidal periodicities which also may be theoretically
related to increased seismicity.
Other defined periodicities may be directly related to solar-
terrestrial seismicity increased such as regular changes in flare
production on the sun. Ichimoto et al. (1985) and Rieger et al.
(1984) have found flare periodicities at 74, 92, 115, and 155 day
periods, for example. These periodicities are believed to reflect
the necessary time for the solar-magnetic field to build up after
a previous flare occurred near the same location on the sun.

METHODS AND RESULTS

The earth interacts with the solar wind and solar
electromagnetic activity in ways that reflect precursory
electromagnetic activity on faults. Earthquake lights and auroras
may be confused since both may precede increased seismic activity;
earth currents may appear to be generated by faults in the final
steps of earthquake preparation, but may, instead, be induced by
ionospheric currents triggered by solar storms; EME may be related
to fault activity, to ionospheric activity, to solar radio
activity, or to passing seismic waves. There is abundant
circumstantial evidence that a triggering relation between solar-
induced electromagnetic effects can trigger earthquakes in
California. For example, the Loma Prieta (Ms 7.2, Oct 18,1989) and
the Morgan Hill (Ms 6.2, May 2, 1983) occurred on the same day as
two of the largest flares in the past 20 years. Both the Landers
(Ms 7.5, 6-18-1992) and the Northridge earthquake (Ms 6.7, 1-17-
1994) occurred within several days of flares which were the largest
within 6 months of the their occurrence. Elsewhere, during the last
solar active period (1989-1991) each of the largest five solar
flares was followed by an earthquake of Ms>=6.0 in the African Rift
valley. More recently, each geomagnetic storm of AA>=40 during the
period 1997-1999 has been followed 4-6 days later by a Mb>=5.0
quake in southern China or northern India.
My approach is to ignore long-term variations and concentrate
on short term correlations on a regional and global basis. To do
this, I have split the NOAA earthquake catalog (Whiteside et al.
1996) into a number of regional and local catalogs which were then
cross-correlated in time with several measures of solar and
geomagnetic activity. The number of sunspots is a measure of the
area of the sun which is active at a particular time. This is the
index used in most previous studies since it has been kept
consistently since the late 18th century. A second index, solar
flares are taken from the NOAA catalog. This catalog is fairly good
from about 1918, but is easier to use and more complete since 1968.
The last index is the AA* index available from NOAA through the
internet for the time period 1868-1998. The AA* index measures the
stability of the geomagnetic field. It is lowest when the
geomagnetic field is stable, highest with high geomagnetic field
fluctuation. Other indices which have been used are the DST index -
a measure of the strength of the equatorial ring currents and the
AE index, measuring the strength of the auroral electrojets (or
ring currents).
For solar effects, I have correlated with earthquakes of on a
daily basis with the AA* index on days which the index value is
greater than 80. These are considered major geomagnetic storms.
Flares of class M5.0 or greater were considered possible triggering
events. The largest 250 to 30 earthquakes in each seismic area were
chosen on the basis of their Ms magnitude. These data covered the
entire length of the solar data and allowed statistical analysis of
each seismic area. However, magnitude limits varied considerably
from area to area.
All interevent times between earthquakes and AA* index,
sunspots or flares were tabulated and counted in one-day intervals.
This produces a relation between number of earthquakes per day vs
time of solar/geophysical event for each of 143 earthquake zones
which were investigated.
For each areas, the numbers of earthquakes per day were
analyzed using the z-statistic. This is a simple statistic which is
applicable when the number of samples is large and a sample
deviation can be determined. Statistical significance levels are
well established for the z-statistic.

Here we may want to show the global and north-American results
(maps) and also the daily counts for a number of areas including
Mammoth Lakes.

I have chosen to examine the roles of SFE in three data sets.
First the Aleutian subduction zone is chosen for examination of
longer term effects. This region has a high seismicity rate and
some extensive aftershock sequences, areal coverage by
seismographic stations is excellent (the ADAK local network) and
teleseismic catalogs are available from NGDC. The region is also in
a high latitude (50-55 N), a factor which may be important in some
solar effects related to charged particles and induced telluric
currents. Other regions studied are the African rift Valley, the
west coast of the U.S. and the African Rift valley.

The Short-Term (0-200 minutes) Effects of SFE on Aftershock
sequences:

Theoretically, there are two types of SFEs which may be seen
in data sets using aftershocks during the first 200 minutes after
the mainshock. First, triggering by solar radio signals, and
second, triggering by causes related to continuous torsional free-
earth oscillations, those induced by the mainshock or possibly
those induced from SFE.
Eight days of aftershocks from the Imperial Valley earthquake
of 15 October, 1979 were used to test for triggering of solar radio
signals. A priori it is not known what type of triggering, if any
should be expected. We use the following procedure. The recorded
minute of initiation and maxima of each outstanding solar radio
emission listed in the Solar-Geophysical Data comprehensive reports
(NOAA, 1980) are chosen as times of initial triggering. The
intensity of each event is related to the number of stations
reporting it, so each event is weighted by the number of observing
stations. During the 8-day aftershock period studied (Oct. 15-23),
888 such observatory reports were made. During the same time
interval there are 1269 listed aftershocks. The time delay from
each radio observation to each of the 1269 aftershocks is found and
compiled in 1-minute duration bins. If there is a particular time
between radio bursts and aftershocks in this sequence it should
show as a peak when time delay is plotted against number of events
at each time delay. This is shown in Figure XX. The background is
non-uniform with peaks occurring at regular intervals with positive
time difference. The strongest peak of nearly +6 standard
deviations from the mean falls between 0 and 1 minute after
outstanding radio bursts. The cross-correlation of the time series
of starts and maxima of radio bursts associated with solar flares
and the time series of Imperial Valley aftershocks (Figure XX)
shows high correlation with a strong central peak and stable
background for the time within 1 day of the flare occurrence.
Magnitude signatures (Habermann, 1987) compare the magnitudes of
events occurring in the first three minutes after radio bursts
associated with solar flares and each other 3-minute period during
the 400 minutes investigated. All investigated magnitude signatures
had the same form, two examples of which are shown in Figure XX.
The negative z-statistic on the "and below" (all events below the
magnitude listed for the foreground and background periods) is
indicative of a two to three-fold increase in the 3 minutes after
solar radio bursts in the number of aftershocks compared with the
three minutes before the radio signal. The higher z-statistic on
the "and above" side suggests a magnitude shift has also occurred
and can be modelled as a shift of 0.5 magnitude units increase in
the 3 minutes after a radio burst over all other periods
investigated. No limits on magnitude were used and listed
magnitudes were from 0.0 to 6.6 Ml. Generally, in the use of
magnitude signatures, it is necessary to remove artificial effects
introduced by changing networks or environmental factors
(Habermann, 1987). The network is unlikely to change in a 3-minute
period and the only environmental factor which is likely to be
involved here is the radio band effect on the seismograph. Since
seismographs are insulated from this type of effect, the increase
in magnitude appears to be real.
Additional short term effects expected from SFE are induced
torsional free-earth oscillations. Recent work (XXX) has shown that
free-earth oscillations occur continuously, even in the absence of
large earthquakes, although they are enhanced following great
earthquakes. The cause of these continuous oscillations is under
debate, however, we hypothesize that some of this activity is
triggered or enhanced by the torsional and spheroidal effects of
the compression and electrification of the ionosphere. The
strongest torsional free-earth oscillations have periods of 43.5,
21.6 and 28 to 29 minutes (Stein and Geller, 1978). Periodograms of
times of aftershocks in Imperial Valley after all flares listed in
NOAA data reports (1980) were created using standard techniques of
fourier analysis and are shown in Figures XX and XX. These show the
same periodograms but with vertical lines at different
periodicities - the 43.5/21.6 minute periodicity is shown in Figure
XX, while the 29.5 and its half 14.75 minute periodicity is shown
in Figure XX. The peak periodicity best corresponds with free
oscillation 3T0 (Figure XX), but is slightly misplaced for 2T0 (43
or 21 minutes). This is likely due to the splitting of the modes
into several different periods the strongest of which are separated
by about 1 minute. Note that in Figure XX the first and second free
spheroidal oscillations at 36 and 53 minutes are absent. Figure XXX
shows the results of the analysis for aftershocks of the Coalinga
earthquake (2 May, 1983). The 43/21 minute periodicity is present,
but at a reduced amplitude from that seen in Imperial Valley. Since
there are nearly twice as many aftershocks in the Coalinga sequence
(2350 events), this reduced amplitude is not the results of less
events, but may be related to the triggering mechanism and its
effect on two different stress and fault regimes.