Welcome to Our Science Section

Forecasting Techniques

There are three distinct categories of earthquake forecasting and warning based on time frame. Long term forecasting addresses the probability of an event within decades; immediate warning provides notice that the earth has actually started shaking; short-term forecasts will provide notice that an event is likely within days or weeks.

Strides have been made in long-term seismic forecasts by studying the history of earthquakes and the structure of the earth’s crust to gauge the probability of an event in the next ten or thirty years. More recently, systems to provide immediate warning have become operational in Japan and have been proposed for California.

Unfortunately short-term forecasts have so far eluded the scientific community, and it is these forecasts that will be most effective at saving lives. However, science has made huge strides in the forecasting of other physical phenomena such as hurricanes and tornadoes. Since earthquakes are physical phenomena, we expect that science will find predicitive qualities on which to base short-term forecasts of seismic activity.

The most promising area of current research is seismo-electromagnetic science. Researchers in this field montitor and analyze subtle effects in the earth and ionosphere that are measurable several hours to several days before major earthquakes. The range of electromagnetic signals observed includes IR “blooms”, visible earthquake lights, ULF magnetic field changes or magnetic unipolar pulses, positive and negative air conductivity changes, ionospheric disturbances (Total Electron Content-TEC changes), sudden radio noise across multiple bands, radio wave propagation changes, and possibly abnormal animal behavior prior to large earthquakes.

Better understanding of these effects may lead to useful forecasts. For that to happen, much more study will be needed, including placement of more instruments in more places,analysis of signals related to more earthquakes, and construction of sophisticated software models.


Although there is no consensus on the dominant physical mechanism responsible for earthquake precursory signals, one way to explain them may be:

  • Rocks near the hypocenter of the impending quake are stressed to their elastic limit and begin to crack –without actually displacing (rupturing) yet.
  • The cracking process releases a flood of charged particles (some researchers say electrons, some say positive charge carriers called p-holes, and some simply say ionic water migrates through the cracks).
  • Moving charges form huge underground currents (10^6 Amps estimated in the 1999 Chi-Chi, Taiwan quake) which disturb the Earth’s normal magnetic field.
  • These disturbances can be detected at ultra-low frequencies (ULF) due to the signal’s ability to penetrate kilometers of solid rock only at low frequencies (electromagnetic skin effect).
  • Some lab experiments have also shown that p-holes can migrate to the surface, drop their charges, and emit IR radiation in discreet frequencies bands (NASA-Freund) that have been detected by satellite IR instruments (NASA-Ouzounov).

During the past decade the source of these signals has been the subject of much investigation and several theories have emerged:

  • Semiconductor effect [Freund, 2000]
  • Piezomagnetism [Johnston, 1997]
  • Resistivity changes: [Park et al., 1993]
  • Streaming potential: [Fenoglia et al., 1995]
  • Magnetohydrodynamic [Molchanov et al., 2001]
  • Piezoelectric effect [Finkelstein et al., 1973]
  • Triboelectricity [Gokhberg et al., 1982]
  • Fluid vaporization [Chalmers, 1976]

Summary of prominent theories

Semiconductor effect [Freund, 2000]: when rock starts to fracture, charge carriers (p-holes), are released, which in turn generate large ground currents and corresponding magnetic field disturbances, as well as IR signatures (in 8 and 11m) bands when the charges are neutralized on the earth’s surface.

Piezomagnetism: the magnetic properties of rock are known to change as a function of applied stress [Revol et al., 1977] and have been combined with stress changes from dislocation models of fault rupture [Johnston, 1978; Banks et al., 1991] to produce magnetic field fluctuations of a few nanoTesla, which are readily observed [Johnston, 1997].

Resistivity changes: the electrical resistivity of porous rocks is known to change as a function of compression and shear [Brace et al, 1965; Yamazaki, 1965], which can be measured using passive (telluric or magnetotelluric) or active experiments [Park et al., 1993].

Streaming potential: when a conducting fluid is forced to flow past a stationary charged surface, an electrokinetic effect can cause currents to begin flowing either in the fluid itself or in the surrounding rock [Fitterman 1979], which, under realistic crustal parameters can generate surface magnetic fields of a few nanoTesla [Fenoglia et al., 1995].

Magnetic signal generation by either conducting fluid flowing in the presence of the Earth’s magnetic field or by the magnetohydrodynamic conversion of a seismic signal into an electromagnetic signal during propagation in a conductive medium [Molchanov et al., 2001]: While these mechanisms have been proven to work in laboratory settings, it is unclear how they behave in conditions more closely approximating those of the core (wet rocks under confining pressure) due to the absence of detailed measurements in active fault zones.

Other proposed mechanisms include piezoelectric effect [Finkelstein et al., 1973], triboelectricity [Gokhberg et al., 1982], and fluid vaporization [Chalmers, 1976].