based on the lecture notes of Prof. Stephen A. Nelson - Tulane University
Long-term forecasting is based mainly on the knowledge of when and where
earthquakes have occurred in the past. Thus, knowledge of present
tectonic setting, historical records, and geological records are studied
to determine locations and recurrence intervals of earthquakes. Two
aspects of this are important.
- the study of prehistoric earthquakes. Through study of the
offsets in sedimentary layers near fault zones, it is often possible to
determine recurrence intervals of major earthquakes prior to the recorded
historical record. If it is determined that earthquakes have
recurrence intervals of say 1 every 100 years, and there are no records of
earthquakes in the last 100 years, then a long-term forecast can be made
and efforts can be undertaken to reduce seismic risk.
Example: The diagram below shows a hypothetical cross-section of a
valley along a fault zone. The valley has been filled over the years
with clays, sands, and peat (decaying organic matter). The upper
peat layer is not yet cut by the fault. Peat is a useful material to
geologists, since it contains high amounts of Carbon that can be dated
using the 14C method. The ages for each of the peat layers are
shown. The dates suggest that a major faulting event cut the lower
peat layer sometime after it was deposited 440 years ago. The dates
also show the middle peat layer was cut by a faulting event after it was
deposited 300 years ago. If these faulting events were associated with
earthquakes, this suggests a recurrence interval of about 140
years. Since the upper peat layer has not yet been cut by the fault
and is 135 years old, we can speculate that within the next 10 years or so
there may be another earthquake. This assumes, of course, that the
two previous events are an accurate measure of the recurrence interval.
Seismic gaps - A seismic gap is a zone along a tectonically active area
where no earthquakes have occurred recently, but it is known that elastic
strain is building in the rocks. If a seismic gap can be
identified, then it might be an area expected to have a large earthquake
in the near future.
Example - The Mexico Earthquake of 1985
The map below shows the southern coast of Mexico. Here the Cocos
plate is subducting beneath the North American Plate along the Acapulco
Trench. Prior to the September of 1985 it was recognized that
within recent time there had been major and minor earthquakes on the
subduction zone in a cluster pattern. For example, there were
clusters of earthquakes around a zone that included a major earthquake on
Jan 30, 1973, another cluster around an earthquake of March 14, 1979, and
two more cluster around earthquakes of July 1957 and January, 1962.
Between these clusters were large areas that had produced no recent
earthquake activity. The zones with low seismically are called
seismic gaps. Because the faulting had occurred at other places
along the subduction zone it could be assumed that strain was building in
the seismic gaps, and earthquake would be likely in such a gap within the
near future. Following a magnitude 8.1 earthquake on September 19,
1985, a magnitude 7.5 aftershock on Sept. 21, and a magnitude 7.3
aftershock on Oct. 25, along with thousands of other smaller aftershocks,
the Michoacan Seismic gap was mostly filled in. Note that there
still exists a gap shown as the Guerrero Gap and another farther to the
southeast. Over the next 5 to ten years we may expect to see
earthquakes in these gaps.
Example - The San Francisco, Loma Prieta, and Parkfield Seismic Gaps
Shown below are two cross-sections along the San Andreas Fault in northern
California. The upper cross section shows earthquakes that occurred
along the fault prior to October 17, 1989. Three seismic gaps are
seen, where the density of earthquakes appears to be lower than along
sections of the fault outside the gaps. To the southeast of San
Francisco is the San Francisco Gap, followed by the Loma Prieta Gap, and
the Parkfield Gap. Because of the low density of density of earthquakes in
these gaps, the fault is often said to be locked along these areas, and
thus strain must be building. This led scientist to issue a
prediction for the Parkfield gap that sometime between 1986 and 1993 there
would be an earthquake of magnitude 6 or greater south of Parkfield.
No such earthquake has yet occurred. However a magnitude 7.1
earthquake occurred in the Loma Prieta gap on Oct. 17, 1989, followed by
numerous aftershocks. Note how in the lower cross-section, this
earthquake and its aftershocks have filled in the Loma Prieta Gap.
This still leaves the San Francisco and Parkfield gaps as areas where we
might predict a future large event.
Short-term predication involves monitoring of processes that occur
in the vicinity of earthquake prone faults for activity that signify a
Anomalous events or processes that may precede an earthquake are
called precursor events and might signal a coming earthquake.
Despite the array of possible precursor events that are possible to
monitor, successful short-term earthquake prediction has so far been
difficult to obtain. This is likely because:
the processes that cause earthquakes occur deep beneath the
surface and are difficult to monitor.
earthquakes in different regions or along different faults all
behave differently, thus no consistent patterns have so far been
Among the precursor events that may be important are the following:
Ground Uplift and Tilting of the ground - Measurements taken in the
vicinity of active faults sometimes show that prior to an earthquake
the ground is uplifted or tilts due to the swelling of rocks caused by
strain building on the fault. This may lead to the formation of
numerous small cracks (called microcracks). This cracking in
the rocks may lead to small earthquakes called foreshocks.
Foreshocks - Prior to a 1975 earthquake in China, the observation
of numerous foreshocks led to successful prediction of an earthquake
and evacuation of the city of the Haicheng. The magnitude 7.3
earthquake that occurred, destroyed half of the city of about 100
million inhabitants, but resulted in only a few hundred deaths because
of the successful evacuation..
Water Level in Wells - As rocks become strained in the vicinity of a
fault, changes in pressure of the groundwater (water existing in the
pore spaces and fractures in rocks) occur. This may force the
groundwater to move to higher or lower elevations, causing changes in
the water levels in wells.
Emission of Radon Gas - Radon is an inert gas that is produced by
the radioactive decay of uranium and other elements in rocks.
Because Radon is inert, it does not combine with other elements to
form compounds, and thus remains in a crystal structure until some
event forces it out. Deformation resulting from strain may force
the Radon out and lead to emissions of Radon that show up in well
water. The newly formed microcracks discussed above could serve
as pathways for the Radon to escape into groundwater. It has
been reported that increases in the amount of radon emissions
increases prior to some earthquakes.
Changes in the Electrical Resistivity of Rocks - Electrical
resistivity is the resistance to the flow of electric current .
In general rocks are poor conductors of electricity, but water is more
efficient a conducting electricity. If microcracks develop and
groundwater is forced into the cracks, this may cause the electrical
resistivity to decrease (causing the electrical conductivity to
increase). In some cases a 5-10% drop in electrical resistivity
has been observed prior to an earthquake.
Unusual Radio Waves - Just prior to the Loma Prieta Earthquake of
1989, some researchers reported observing unusual radio waves.
Where these were generated and why, is not yet known, but research is
Strange Animal Behavior - Prior to a magnitude 7.4 earthquake in
Tanjin, China, zookeepers reported unusual animal behavior.
Snakes refusing to go into their holes, swans refusing to go near
water, pandas screaming, etc. This was the first systematic
study of this phenomenon prior to an earthquake. Although other
attempts have been made to repeat a prediction based on animal
behavior, there have been no other successful predictions.
Controlling Earthquakes Although no attempts have yet been made to control
earthquakes, earthquakes have been known to be induced by human
interaction with the Earth. This suggests that in the future
earthquake control may be possible.
Examples of human induced earthquakes
For ten years after construction of the Hoover Dam in Nevada
blocking the Colorado River to produce Lake Mead, over 600 earthquakes
occurred, one with magnitude of 5 and 2 with magnitudes of 4.
In the late 1960s toxic waste injected into hazardous waste disposal
wells at Rocky Flats, near Denver apparently caused earthquakes to
occur in a previously earthquake quiet area. The focal depths of
the quakes ranged between 4 and 8 km, just below the 3.8 km-deep
Nuclear testing in Nevada set off thousands of aftershocks after the
explosion of a 6.3 magnitude equivalent underground nuclear
test. The largest aftershocks were about magnitude 5.
In the first two examples the increased seismicity was apparently due to
increasing fluid pressure in the rocks which resulted in re-activating
older faults by increasing strain.
The problem, however, is that of the energy involved. Remember that
for every increase in earthquake magnitude there is about a 30 fold
increase in the amount of energy released. Thus, in order to release
the same amount of energy as a magnitude 8 earthquake, 30 magnitude 7
earthquakes would be required. Since magnitude 7 earthquakes are
still very destructive, we might consider generating smaller earthquakes.
If we say that a magnitude 4 earthquake might be acceptable, how
many magnitude 4 earthquakes are required to release the same amount of
energy as a magnitude 8 earthquake? Answer 30 x 30 x 30 x 30 x 30 =
810,000! Still, in the future it may be possible to control
earthquakes either with explosions to gradually reduce the stress or by
pumping fluids into the ground.