Expert Commentary

Test of forecasting model on Japanese earthquakes

2005 study in Pure and Applied Geophysics testing an earthquake forecasting model against the records of Japanese quakes over 36 years.

If ever there were a natural event worth forecasting, it would be a major earthquake. Currently, there are dozens of long-term predictive models employed around the world that attempt to do just that with varying degrees of success. One such model, named EEPAS (Every Earthquake a Precursor According to Scale), attempts to predict the magnitude, location and probability of a future major quake based on indicative measures of the minor earthquakes that precede it.

A 2005 study published in Pure and Applied Geophysics, “Test of the EEPAS Forecasting Model on the Japan Earthquake Catalogue,” tested the accuracy of this predictive system on the pattern of natural phenomena between 1965 and 2001.

The findings included:

  • When tested alongside a popular baseline forecasting model (known as Proximity to Past Earthquakes or PPE) the EEPAS model (using generalized parameters) modeled Japan’s historic earthquake activity more accurately.
  • The generalized EEPAS model was even more accurate when fitted precisely with parameters sensitive to the Japanese environment. This indicates that models purporting to predict activity across different seismic regions may not perform as well as region-specific models.

Although EEPAS is just one of many theoretical models for forecasting the magnitude and location of future earthquakes, the study’s results indicate that continued research is important. As Japan’s 2011 earthquake and subsequent Fukushima nuclear disaster indicate, the devastation caused by such events, both in human and economic terms, can be immense.

Tags: California, disasters, safety, Asia

About The Author