from the book
Earthquake Prediction From Seismic Electrical Signals
Sir James Lighthill
University College London


Earthquake Prediction Research Center, Tokai University. Shimizu 424, Japan and Geodynamics Research Institute/Department of Geology and Geophysics, Texas A&M University, College Station, Texas 77843, USA

Earthquake prediction has been in practice in Greece for more than a decade by the so called VAN method. The method is based on detection of characteristic changes in the geoelectric potential, the so-called Seismic Electric Signals (SES) that appear prior to earthquakes. SES are distinguished from noise through a set of criteria based on simple physical principles. It has been found that SES are observed only at particular locations (sensitive sites) and that a sensitive site is selectively sensitive to SES from particular seismic source area's), making epicentral prediction possible to within about l00 km. Magnitude of an impending earthquake is predicted, to within 0.7 units, from its relationship, also empirically discovered, with epicentral distance and intensity of SES.
The main published counter-arguments against the effectiveness of VAN are these:
1) SES are noise, 2) the success and alarm rates are not as high as the VAN group claims, 3) such a success rate can be achieved by chance and 4) The VAN method lacks physical mechanism. Some facts and thoughts on these objections are presented.

In the 1970's, optimism prevailed in geoscientific community that successful earthquake prediction was in sight. The gap strategy, Haisheng prediction, and the dilatancy model, among others, contributed to that optimism. However, subsequent events yielded only discouraging results, so that the present consensus appears to be that earthquake prediction, in particular short-term, prediction, is still beyond the foreseeable future. Some theoretical seismologists. even believe that earthquake prediction would, be impossible in principle because of intrinsic chaotic nature of fracture process . However, there is an outstanding exception to the general pessimism, which is the success of the VAN method in Greece. The VAN method is unique in that it has been successfully predicting earthquakes for more than a, decade. Naturally, the VAN method has been highly controversial. Some of the criticism and confusion, however, seems to be rooted on misunderstanding. It is the intent of this paper to provide a general introduction to the VAN method and to try to help removing misunderstanding.

Thanks to seismology, we now know that earthquakes are caused by sudden fault displacements, which in turn are due to stress accumulated at plate boundaries and in plate interiors by plate motions. However, their prediction is still very difficult. For earthquake prediction, one has to specify three elements, i.e., when, where and how large the impending earthquake will be, with precision useful for human society. With regard to "when", it is customary to classify prediction into three categories, namely the long-term (more than tens of years), intermediate-term (one to lens of years) and short-term (less than one year) prediction. 

Former two classes of predictions, mainly based on the past history of fault movement and seismicity, can be useful for long-term planning for mitigation of earthquake hazard. On the other hand, short-term prediction is based on precursory phenomena immediately before earthquakes. A wide variety of phenomena, including seismological, geodetic, geochemical, hydrological, electro-magnetic or even meteorological and biological phenomena, has been postulated to be potential precursors. Except for a few cases like Haisheng prediction, however, there has practically been no success but frequent failures. Although some success may be hoped in the future, owing to the development of new technologies and data handling techniques, such as GPS, satellite interferometry, water table monitoring, so far, at best, most of the reported precursory phenomena have been noticed only after earthquakes. They are called post-predictions.

The VAN method, named after the initials of Professors P. Varotsos, K. Alexopoulos and K. Nomicos, stands out as a notable exception in that it has been actually making short-term predictions before, not after, earthquakes. Soon after the disastrous earthquake in Athens area in 1981, the VAN group started monitoring geoelectric potential changes, because solid state physicists Varotsos and Alexopoulos anticipated theoretically that some electric current would be generated in the earthquake source region just prior to earthquake. Nomicos developed the necessary data acquisition system. The VAN group now claims that earthquakes in Greece with Ms(ATH) (magnitude announced by the Seismological Institute of the National Observatory of Athens, SI-NOA) greater than 5 can be predicted within the error of 100 km in epicentral location and 0.7 unit in magnitude. The time of earthquake occurrence is claimed to be from several hours to II days after detecting the signals, but it can be several weeks for repeated and prolonged signals.

A Dislocation Model For Seismic Electric Signals
Notes on Generation and Propogation of Seismic Transient Electric Signals

Theoretical Aspects and References
Main Page