For a statement to be accepted as a valid earthquake prediction, it has to contain the expected magnitude with error limits, the well defined area of the epicenter, the range of dates, and the probability of this to come true. The data from which the prediction was derived must be verifiable and the analysis of these data must be reproducible. Long term predictions (years to decades) are more likely to be achieved than medium term predictions (months to years), and short term predictions (hours to days) are in general unlikely to be possible, at present. If a plausible mechanism linking the observations with the predicted earthquake is not offered, the credibility of the prediction is diminished, but it may not necessarily be rejected. Evaluations of apparent successes must include a statistical estimate of the probability that the prediction came true by chance, which is often the case with predictions by amateurs. Whether a prediction is scientific or amateurish is not based on who makes the prediction, but based on how the prediction is made and tested. Predictions can be formulated either by defining the limits of the parameters probabilistically or by firm values.
The problem of earthquake prediction
In efforts to predict earthquakes, seismologists have investigated the association of an impending earthquake with such varied phenomena as seismicity patterns,[1][2][3][4][5][6] crustal movements,[7][8] ground water level in wells,[9] radon or hydrogen gas emissions from the Earth,[10][11] changes of seismic wave velocities,[12] electromagnetic fields (seismo-electromagnetics),[13] large-scale changes in soil temperature,[14] and changes in ion concentration in the ionosphere.[14] A relatively recent review is found in [15] and a special issue of Pure and Applied Geophysics contains a collection of recent articles on the subject.[16]
The mystery of earthquake occurrence frequently sparks people without scientific training into claiming that they have found the solution to the earthquake prediction problem. Discredited, fantastic theories of predicting earthquakes include weather conditions and unusual clouds, and the phases of the moon.[17] These pseudoscientific theories and predictions[18] ignore the requirement of rigorously formulating the hypothesis and to test it statistically [1].
Self-appointed prediction experts often resort to the technique of making vague statements, which they claim were correct predictions, after an earthquake has happened somewhere.[19] Rudolf Falb's "lunisolar flood theory" is a typical example from the late 19th century.
Precursory seismicity patterns
[edit] The Mogi doughnut
In 1969 Japanese seismologist Kiyoo Mogi proposed what has become known as the 'Mogi doughnut hypothesis', which suggests that major earthquakes tend to occur in an unusually seismically calm area surrounded by a ring of unusually high seismic activity.[20][21][22]
[edit] Geochemical precursors
[edit] Radon
Emission of radon as a quake precursor was studied in the 1970s and 80s with no reliable results and continued to be dismissed by most seismologists until recently. However, after the 2009 L'Aquila earthquake, which was preceded by an Italian laboratory technician's predictions of an impending major earthquake, some in the scientific community expressed renewed interest in radon as a quake precursor. In December 2009, the technician, Giampaolo Giuliani, presented his research to the American Geophysical Union in San Francisco and was later invited by the American Geophysical Union to participate in developing a worldwide seismic early warning system.[23]
Further information: Radon#Scientific
[edit] Precursory crustal deformation
Under construction
[edit] Electromagnetic precursors
[edit] VAN claims
Main article: VAN method
VAN is a controversial method of earthquake prediction proposed by Professors Varotsos, Alexopoulos and Nomicos in the 1980s; it was named after the researchers' initials. The method is based on the detection of "seismic electric signals" (SES) via a telemetric network of conductive metal rods inserted in the ground; it stems from theoretical predictions by P. Varotsos, a solid-state physicist at the National and Capodistrian University of Athens.[24] Researchers have claimed to be able to predict earthquakes of magnitude larger than 5, within 100 km of epicentral location, within 0.7 units of magnitude and in a 2-hour to 11-day time window, but this is disputed.
[edit] Combined measurements of different precursors
Under construction
[edit] Foreshocks
Foreshocks are medium-sized earthquakes that precede major quakes.
An increase in foreshock activity[25] (combined with purported indications like ground water levels and strange animal behavior) enabled the successful evacuation of a million people one day before the February 4, 1975 M7.3 Haicheng earthquake[26] by the China State Seismological Bureau.
While 50% of major earthquakes are preceded by foreshocks, only about 5-10% of small earthquakes turn out to be foreshocks, leading to false warnings.[25][27][28]
[edit] Earthquake forecasts and partial predictions
[edit] Early warning
An earthquake warning system is a system of accelerometers, communication, computers, and alarms that is devised for regional notification of a substantial earthquake while it is in progress. Japan, Taiwan and Mexico all have earthquake early-warning systems.
In a paper in the journal Nature, Richard Allen of the University of California claims that the distinction between small and large earthquakes can be made from the very first seconds of seismic energy recorded by seismometers, though other scientists are not convinced.[29] If correct this may make earthquake early warning (as distinct from prediction) more powerful. Earthquake early warning provides an alarm that strong shaking is due soon to arrive, and the more quickly that the magnitude of an earthquake can be estimated, the more useful is the early warning. However, earthquake early warning can still be effective without the ability to infer the magnitude of an earthquake in its initial second or two.
[edit] Triggering earthquakes
[edit] Solid Earth tides
There are two flavors of tidal stressing that have been claimed to generate enhanced rates of earthquakes—diurnal and biweekly tides. The diurnal correlations would arise from more earthquakes only during the hours when the tidal stress is pushing in an encouraging direction, in contrast, biweekly effects would be based on earthquakes occurring during the days when the sinusoidal stressing oscillations are largest. The former, as most easily observed in the twice-daily rise and fall of the ocean tides, have occasionally been shown to influence earthquakes (e.g.,[30] this paper shows there may be some weak tidal triggering of shallow, oceanic thrust-faulting earthquakes). The latter, which arises from the periodic alignment of the Sun and Moon, has often been claimed in the popular press to incubate earthquakes (sometimes termed the "syzygy" effect) and occasionally for small datasets in the scientific literature (e.g.,[31]), but generally such effects do not appear in careful studies of large datasets.
A paper published in Taiwan, by the Department of Astronomy, Beijing Normal University, found a significant relationship to tidal forces and earthquakes in China and Taiwan. The paper considers the relationship between 21 major earthquakes (Ms ≥ 7.0) in land and the offshore area of Taiwan island in the 20th century and the variance ratio of the lunar-solar tidal force. The result indicates that the time of these earthquakes is closely related to the variance ratio of the lunar-solar tidal force, and therefore that the tidal force possibly plays an important role in triggering earthquakes.[32] The conclusion is this method may be used to help forecast earthquakes by studying the lunar perigee.
Syzygy, which is not given much credence in the scientific community, is motivated by the observation that, historically, there have been some great earthquakes whose timing coincides with tidal forces near their maximum. For maximum tidal force, three factors must coincide: first, when the moon (in its elliptical orbit) is closest to the earth; second, when it is within a day or two of a new moon (so that the tidal forces of the moon and sun are acting in concert); and third, when the earth (in its elliptical orbit) is at or near its closest distance to the sun.
Shallow earthquakes near mid-ocean ridges, volcanic earthquakes, and episodic tremor and slip have also been observed to sometimes correlate with the diurnal tides, with enhanced activity correlating with times that faults are unclamped.
[edit] Evaluation of prediction claims and methods
[edit] National prediction evaluation councils
Official earthquake prediction evaluation councils have been established in California (the California Earthquake Prediction Evaluation Council) and the federal government in the United States (the National Earthquake Prediction Evaluation Council), but have yet to endorse any method of predicting quakes as reliable.[25]
Unless the following parameters are specified, a statement does not qualify as an earthquake prediction:[33]
A specific location or area
A specific span of time
A specific magnitude range
A specific probability of occurrence
[edit] International evaluation of claims and methods
The sub-commission for earthquake prediction of IASPEI (International Association of Seismology and Physics of the Earth’s Interior) has reviewed claims of successful predictions and of proposed methods to predict during the 1990s. Their procedure was similar to reviews of proposals for research grants. Authors submitted their detailed research on the prediction problem. Anonymous reviewers commented, and members of the sub-commission discussed the merits of the proposal and of the reviewer’s comments.
A decision to place the claim into one of three categories (preliminary list of significant precursors, no decision, rejected) was then transmitted to the authors, who could write a reply, if they so wished. The entire exchange was then published, unless the authors did not agree to publication.[34][35] Most of the nominated successful predictions and methods to predict were rejected.[36] At that time, three methods seemed most promising: Seismicity patterns, ground water properties, and crustal deformations.[37]
Attribution to a plausible physical mechanism lends credibility, and suggests a means for future improvement. Reproducibility and statistical analysis are used to distinguish predictions which come true due to random chance (of which a certain number are expected) versus those that have more useful predictive capability, and to validate models of long-term probability. Such models are difficult to test or validate because large earthquakes are so rare, and because earthquake activity is naturally clustered in space and time. "Predictions" which are made only after the fact are common but generally discounted.
[edit] The problem of anecdotal claims
[edit] Animal behavior
Animal behavior reports are often ambiguous and not consistently observed. In folklore, some animals have been identified as being more able to predict earthquakes than others, especially dogs, cats, chickens, horses, toads and other smaller animals.
It has been postulated that the reported animal behavior before an earthquake is simply their response to an increase in low-frequency electromagnetic signals.[38] The University of Colorado has demonstrated that electromagnetic activity can be generated by the fracturing of crystalline rock. Such activity occurs in fault lines before earthquakes. According to one study, electromagnetic sensors yield statistically valid results in predicting earthquakes.[39]
In Italy, findings from 2009 suggest that toads are able to detect pre-seismic cues.[40]
[edit] Fractoluminescence
One possible method for predicting earthquakes, although it has not yet been applied, is fractoluminescence. Studies at the Chugoku National Industrial Research Institute by Yoshizo Kawaguchi have shown that upon fracturing, silica releases red and blue light for a period of about 100 milliseconds. Kawaguchi attributed this to the relaxation of the free bonds and unstable oxygen atoms that are left when the silicon oxygen bonds have broken due to the stresses within the rock.[41]
[edit] Satellite observations
[edit] Demeter microsatellite
The "Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions" satellite, constructed by CNES, has made observations which show strong correlations between certain types of low frequency electromagnetic activity and the most seismically active zones on the Earth, and have shown a sharp signal in the ionospheric electron density and temperature near southern Japan seven days before a 7.1 magnitude occurred there (on August 29 and September 5, 2004, respectively).[42]
[edit] QuakeSat nanosatellite
Quakesat is an earth observation nanosatellite based on 3 CubeSats. It was designed to be a proof-of-concept for collecting extremely low frequency earthquake precursor signals from space. The primary instrument is a magnetometer housed in a 2 foot (0.6 m) telescoping boom.
[edit] The ESPERIA Project
ESPERIA is an equatorial space mission mainly concerned with detecting any tectonic and preseismic related signals. More in general, it has been proposed for defining the near-Earth electromagnetic, plasma, and particle environment, and for studying perturbations and instabilities in the ionosphere-magnetosphere transition region. To study earthquake preparation processes and anthropogenic impacts in the Earth's surface, a phase A study has been realized for the Italian Space Agency.[43]
[edit] DESDynI radar satellite
The Deformation, Ecosystem Structure and Dynamics of Ice (DESDynI) radar satellite, which was canceled in the White House's 2012 budget proposal, would have the capacity to identify elastic strain in tectonic plates, combining L-band interferometric synthetic aperture radar and a multi-beam infrared lidar to detect strains in the Earth’s surface that could lead to serious earthquakes. [44][45]
[edit] TwinSat
Russia and the United Kingdom have agreed to jointly deploy two satellites in 2015 that will measure electromagnetic signals that are released from the earth's crust prior to earthquakes. The project is said to be able to "help predict earthquakes and potentially save thousands of lives." [46]
[edit] History of prediction attempts
[edit] China
After a series of foreshocks, the Chinese government was able to successfully evacuate much of the populace before the 1975 Haicheng earthquake. However, the Chinese government failed to predict the July 28, 1976 M7.8 Tangshan earthquake, which put Chinese earthquake prediction research in doubt for several years. In the late 1990s, there were over thirty false alarms unofficially announced in China,[47] but the Chinese government claimed successful prediction of the November 29, 1999, M5.4 Gushan-Pianling Earthquake in Haicheng city and Xiuyan city, Liaoning Province.[48]
[edit] Japan
The Japanese government established the Imperial Earthquake Investigation Committee in 1892 in response to the Nobi (Mino-Owari) earthquake (1891) which caused significant damage in Japan.[49]
In the 1970s and 1980s, the Japanese government embarked on a major earthquake preparedness campaign, which some criticized as emphasizing prediction too much over mitigation.[50] It failed to result in a prediction of the Great Hanshin earthquake which devastated the city of Kobe in 1995. See also 2011 Tōhoku earthquake and tsunami.
[edit] Failed Lima prediction
An earthquake predicted by a scientist at the U.S. Bureau of Mines to occur on June 28, 1981, in Lima, Peru, failed to materialize. Despite being dismissed by the U.S. National Earthquake Prediction Evaluation Council, the prediction caused popular fear and many left the city.[51]
[edit] Failed Parkfield earthquake prediction
Main article: Parkfield earthquake
Based on a history of regularly spaced earthquakes in the early 20th century, the USGS in 1985 began an experiment based on the predictions and published papers of Allan Lindh and W.H. Bakun of the USGS and T.V. McEvilly of the University of California at Berkeley. The goal was to predict a 6.0 magnitude earthquake near Parkfield, California.[52]
"Bakun and Lindh summarized the state of the art in the Parkfield Prediction Experiment, and predicted that a moderate-size earthquake would occur at Parkfield between 1985 and 1993. Their prediction was unusual both in its precision (as to location, time and magnitude) and high degree of confidence (95% within the 9-year window). Bakun and Lindh (1985) also suggested that the predicted earthquake could produce extended rupture of the San Andreas fault to the southeast, possibly growing to magnitude 6.5 to 7.0."[53]
Media attention focused on the prediction and the experiment. 122,000 pamphlets were mailed to residents of the Parkfield area, entitled "The Parkfield Earthquake Prediction."[54] Despite the prediction, such an earthquake did not occur until after the end of the prediction window, in 2004.[25]
[edit] Loma Prieta prediction
From 1968 to 1988 scientists in California mapped seismic activity on a cross section of the fault lines. They identified a "seismic gap" in the Loma Prieta area from various features of the regional seismicity. They therefore concluded that Loma Prieta was due for an earthquake.[citation needed] Smaller quakes several months beforehand were treated as possible foreshocks, but the warnings had expired by the date of the moment magnitude 6.9 quake, on 17 October 1989.[25]
Further information: 1989 Loma Prieta earthquake
[edit] Failed New Madrid prediction by Iben Browning
In 1989 Iben Browning predicted a major earthquake in the New Madrid fault zone of southern Missouri and specified December 2 or 3, 1990, as the most likely days. This prediction was reported on extensively in the media and lead to great community concern. No earthquake occurred on those days or thereafter.
[edit] Jim Berkland
Jim Berkland claims to have predicted the Loma Prieta quake,[55] but the mainstream scientific community does not endorse his techniques as repeatable, attributing his success with this quake partly to random chance.
[edit] Failed SoCal prediction
In early 2004, a group of scientists at the University of California, Los Angeles, led by Dr. Vladimir Keilis-Borok, predicted that a quake similar in strength to the San Simeon earthquake of 2003 would occur in a 12,000 square mile (31,100 km) area of Southern California by September of that year. The odds were given as 50/50.
In April 2004, the California Earthquake Prediction Evaluation Council (CEPEC) evaluated Keilis-Borok's prediction and reported to the California State Office of Emergency Services.[56] CEPEC concluded that the "uncertainty along with the large geographic area included in the prediction (about 12,400 square miles) leads (us) to conclude that the results do not at this time warrant any special policy actions in California.” The predicted time window came and went with no significant earthquake.
[edit] L'Aquila controversy
Italian technician Giampaolo Giuliani claims to have predicted the 2009 L'Aquila earthquake. He was reported to Italian police for "causing fear" but he was acquitted.[57] His prediction was dismissed by other scientists and politicians as a fluke.[58]
[edit] 2010 Chile earthquake
A group of scientists met with Chilean officials in Talcahuano several months prior to the 2010 Chile earthquake, warning them of the possibility of a large earthquake and tsunami. Daniel Melnick, a geologist from Potsdam University, called this an example of a correct prediction. Government authorities, however, did not take action
