Another Forecast of a Great Earthquake Near Japan

john's picture

Over the weekend I received an email from a colleague, Professor Vladimir ("Volodya") Kossobokov, who works at the Institut de Physique du Globe in Paris.  Professor Kossobokov has been a close colleague of Professor Vladimir Keilis-Borok, who is famous for having initiated studies of pattern recognition applied to earthquakes, US presidential elections (along with Allan Lichtman at the American University in Washington, DC) and other areas of interest to forecasters.

Professor Keilis-Borok is mentioned in Nate Silvers' new book, "The Signal and the Noise", in connection with a prediction of the October 17, 1989 Loma Prieta earthquake.  Some may recall that Soviet General Secretary Gorbachov handed US President Reagan a note at the October 1986 Reykjavik, Iceland, international summit to the effect that a major earthquake was expected in northern California within a few years.  When the Loma Prieta event happened about 150 km south of San Francisco 3 years later, it led to major interest in the Russian methods.

Professor Kossobokov's weekend message to about 200 scientist-colleagues was one of a number of periodic messages that his group sends to those of us interested in earthquake forecasting.  They have been making predictions of epicenters of impending large and great earthquakes for over 10+ years, and sending these out over the internet.  It is their way of carrying out a true prospective test of their methods -- using the internet to place their predictions on record with colleagues.

The Russian epicenter prediction methods are based on a variety of signals.  Among them are small earthquake "chains", together with another method "reverse tracing of precursors".  They have developed several algorithms for prediction, among them a method they call M8.  In this method, they construct many circles of radius 667 km around seismically active regions, and determine whether great earthquakes M>8 are likely to occur in these regions.

I have received permission from Professor Kossobokov to show several of his circles in the Japan region from his recent message, as well as the current global forecast for potential regions as likely candidates for large earthquakes.  Unlike the Open Hazards forecasts, which are stated in terms of probability, the Russian predictions are "alarms", meaning that an alarm is given for a specific region for a specific duration.  Alarms can be turned on and then off, with no earthquake occurring.  So if the earthquake occurs with its epicenter in the defined region within the defined time window, the prediction is considered successful.  If no earthquake occurs in the alarm region, it is a false alarm. There can also be failures to predict.

The interesting conclusion to be drawn from their latest prediction is that they predict a M>8 earthquake also in the region of Japan during the next 3-4 years.  The predicted region overlaps well with the highest-probability region for such an event as discussed in my previous blogs.

To understand the figures below, note that the acronym "TIP" means "Time of Increased Probability" of a great earthquake.  The circles indicate that the epicenter of a great earthquake is expected to fall within the circle.


The Russian forecast is below.



Circles used in the M8 Algorithm near Japan.


Risk Alert