Earthquakes - Prediction

Earthquakes - Prediction


Seismic Gap Theory

Description: The theory that all segments of a fault must move by the same amount to accommodate tectonic stresses. The "Gap" is an area of the fault line identified by a lack of earthquakes for an unusually long time. The theory work by identifying areas more likely to experience an earthquake in the near future. 

Example: Loma Prieta Gap - there was a significant lack of activity in the years leading up to 1989 along the Loma Prieta Section of the San Andreas fault, an earthquake was predicted between 1986 and 93. The Central Kuril Gap, Russia, was identified after a significant investigation into seismic gaps around the pacific rim after the Boxing day Tsunami. A 500Km section had not experienced an earthquake since 1780, but sections to the SW and NE there were major earthquakes in 1915 and 1917. A 275 to 450Km stretch was identified and in November 2006 a M8.3, and in January 2006 an 8.2 earthquake occurred in the defined zone.

Usefulness: Can locate regions at risk, but not time or epicentre. Used to predict the next "big one" in Istanbul on the N. Anatolian Fault

Success: Kuril and Loma Prieta (Partial success), effective for long term prediction, Istanbul in alert

1 of 16

Gas Measurements

Description: Radon is released from rocks deep underground during the early fracturing stage of an earthquake and can be measured dissolved in water in wells.

Example: During the 1993 observation period at Kobe, concentrations of radon were stable at 20Bq/l in a well about 30km NE of the 7.2 Kobe event. By the end of Nov 1994 levels had increased to 60Bq/l. On 7 Jan 1995 a huge increase in radon concentration was observed (to ca. 250Bq/l). These high levels dropped suddenly on 10 Jan, one week before the earthquake. By the time of the earthquake levels had returned to about 30 Bq/l, levels confirmed when the station came back on-line on 22 Jan (monitoring equipment had been damaged by the main shock).

Usefulness: If boreholes/wells are drilled, can be very effective, however it needs specialist equipment. Radon has a half life of 98 hours, so it needs to be regularly monitored to detect accurate levels.

Success: Kobe, possibly the Aquila quake, but this is unsubstantiated

2 of 16

Stress Measurements

Description: Measuring the stress rocks are under using strain gauges and various monitors that detect how stressed a rock is. Build ups of stress are precursors to major events, but it is difficult to measure accurately, and stress is difficult to measure directly. The failed Parkfield experiment did not predict an earthquake using this method, but there have been small successes elsewhere.

Example: Parkfield

Usefulness: Limited case study evidence found - may be updated

Success: Parkfield experiment failed, however, depth measurements are having some success with smaller earthquake predictions in California. 

3 of 16

Water Levels

Description: Water levels are known to fluctuate over weeks to hours before an earthquake. This is due to the compression/contraction of the ground, and the draining and returning of water to the water table, wells and lakes.

Example: In the five years before Tangshan, water levels in the water table varied from 40cm below to 80cm above average. Usefulness: There is not set pattern, variations can be seen in stream flow and groundwater levels up to 400km from the epicentre, making location prediction difficult. Can be useful as an immediate warning to locals if interpreted correctly.

Success: Before the Sichuan earthquake streams were filmed flowing backwards and wells overflowing. China have drilled over 100 research wells to over 1000m deep, and Japan have 93 where temp and depth are monitored. Before the Tangshan quake, 3hours to 4 minutes before wells rapidl rose and overflowed. In general 3 things are observed:

1) A gradual lowering of water levels of a period of months or years
2) An accelerated lowering of water levels (rate often exponential) in the final few months or weeks preceding the earthquake.
3) A “rebound” where water levels begin to increase rapidly in the last few days or hours before the main shock.

4 of 16

Tilt Measurements

Description: Ground surfaces are known to tilt in the months before a quake. It is measured using tiltmeters and used widely in Chile

Example: Before the Chile 1985 quake, a tilit of 13cm over a 20km stretch near the Rapel Reservoir was observed. There are networks of tiltmeters around the country. Deformation is typically less than 1% in California and time is the issue. There was no deformation recorded before the 1964 9.4 event in Alaska

Usefulness: They indicate areas where strain is building up, but do not indicate time

Success: Some, but only to indicate where other investigative tools should be used to predict the timing. More research is needed. Niigata, 1965, a drop of 20cm in ground height was observed just to the north of the epicentre.

5 of 16

Electrical Conductivity

Description: Rock resistivity has been noted to drop 10 to 15% before an earthquake, beginning several months before. Garm, a region in the USSR prone to earthquakes has 5 years worth of records recording a drop in resistivity before an earthquake. Resistivity changes as microcracks in the rocks fill with water and water is a better conductor than rock. 

Example: Garm, 1967 to 72, Russia. 

Usefulness: Useful is resistivity is measured and background "normal levels" are known. Requires specialist equipment

Success: Not yet used to predict an earthquake - if so let me know as i cant find a case study!

6 of 16

Foreshocks and Aftershocks

Description: It is well known that earthquakes are preceded by a pattern of foreshocks, and followed by aftershocks, which can be as big as the initial event. This is different depending on which fault line is in question. Over 40% of moderate to large quakes have foreshocks, and over 70% of >M7. However, this can be minutes to hours, or some consider the 2002 Sumatra quake to be a foreshock to the 2004 Boxing Day Earthquake.

Example: Haicheng is widely accepted to have been predicted largely on the pattern of foreshocks. L'Aquila also demonstrated a significant pattern though warnings were ignored by civil authorities

Usefulness: Useful if the fault line is well understood and there is a significant documented history of the region and te regular pattern - background noise - of the area. Only 5-10% of small earthquakes turn out to be foreshocks

Success: Limited, foreshocks are observed, and an increase in seismic activity is a useful warning, but is limited in how it can be used to predict, as it does not give reliable evidence for timescale

7 of 16

Animal Behaviour

Description: Animals are known to behave strangely, some are supposed to be more receptive than others such as dogs, cats, chickens, horses, toads. There is widespread belief animals detect an increase in low frequency electromagnetic signals which can be generated by the fracturing of hard rocks. This is common just before an earthquake but can also show up as a pattern of foeshocks and aftershocks

Example: Haicheng, 1975 was succesfully predicted, 150000 people were saved and just over 2000 died and under 30 000 were injured. However, there is widespread belief this was also due to foreshocks and other phenomena. However the following year 250000 died in Tangshan.

Usefulness: Often ambiguous, poorly recorded and associated with other pre-cursors. Not consistently observed, but useful in raising a local alarm if bizarre behaviour is interpreted correctly.

Success: Possible partial in Haicheng but not completly attributed to animals, there were notable toad migrations in Sichuan 2008 and Italy 2009

8 of 16

Earthquake Lights

Description: Auroras of pale coloured lights that are emitted before earthquakes, they have been reported in Japan, New Zealand and China, and were successfully captured on film before the SIchuan Earthquake of 2008. They are believed to be caused by the fracturing of quartz which is known to produce light for a few seconds but have been recorded lasting tens of minutes. It is believed to be created by piezoelectricity - electricity generated by pressure, or by tectonic forces distrubing the ionosphere.

Example: L'Aquila 2009. Sichuan 2010, Chile 2010, possible capture during an aftershock to the 2011 Japan earthquake and tsunami

Usefulness: Distance from epicentre varies from local to 400km (Gansu). Not always observed so unreliable, can last seconds to minutes and you would need to understand what they were.

Success: None recorded, though the phenomena was not accepted by seismologists until 1967 when they were recorded during the Matsushiro earthquake swarm in Nagano, Japan between 1965-7.

9 of 16


Description: The study of sedimentary layers and offsets, i.e. fossil earthquakes, can determine recurrence intervals for earthquakes before historical records began, and can add to our understanding of fault lines. If the recurrence interval is calculated at once every 200 years, and there hasn't been an earthquake in 200 years, once might be expected. 

Example: Would have been useful in Haiti, where an earthquake had not occurred in over 200 years, but there was no scientific enquiry in the region

Usefulness: Only useful in areas with a long history of sedimentation as this is where the records will be. 

Success: Has been useful in determining the risk to the Cascadia region by exposing the existence of long frequency high magnitude mega thrust earthquakes in a region previously thought to have low seismic risk. That risk is greatest in the form of a Tsunami that could affect the California, British Columbia, Washington and Oregon coastlines. These regions can now prepare. 

10 of 16

blank card

blank card

11 of 16

blank card

blank card

12 of 16

Satellite Measurements

Description: The Demeter Microsatellite detects low frequency electromagnetic activity and seismically active areas. They have detected signals in the ionospheric electron density and temperate around southern Japan seven days before a 7.1 in 2004. The QuakeSat collects low frequency data, the ESPERIA project dedicated to pre-seismic signal detection. 

Example: There are several satellites that focus on earthquake prediction, they are funded by Russia, the UK, Italy, China and the USA. 

Usefulness: Provides constant monitoring but has only been around for a few years, so not necessarily enough data to be able to make accurate predictions as yet

Success: Japan 2004. 

13 of 16

Successfully Predicted Earthquakes

Haicheng - 1975 - 7.3 - 250 000 people saved, thought there is widespread dispute as to how accurate the prediction was and how it was made. Animal behaviour was certainly a key factor, but the pattern of foreshocks is considered more significant

Xiuyan - 1999 - 5.3 November 28th, a prediction issued for the 28th november to the 8th of december for a magnitude 5.0 to 5.9. No injuries or deaths or panic, and well predicted based on two magnitude 4 earthquakes on November the 8th.

Issues - Prediction or Mitigation? Which saves more lives? Prediction is expensive, and earthquake prone regiosn will always be at risk, mitigation may be more effective and more reliable. 

Not all earthquakes are predictable, it is highly dependant on location, level of development, monitoring status and education of the public. 

There are several predictions in the works, the Istanbul Gap is due, the next Big One in California and a Megathrust on the Cascadia

14 of 16

blank card

blank card

15 of 16

blank card

blank card

16 of 16


No comments have yet been made

Similar Geology resources:

See all Geology resources »