A study on the pattern informatics and its application to earthquake prediction in Taiwan

In this study, we investigate the precursory seismicity pattern by pattern informatics (PI) method and attempt to make the forecast by the PI method every season. The concept of pattern informatics is to describe seismicity change include activation and quiescence by a phase dynamic system. The spatial and temporal evolution of dynamic system can be represented by the phase drift. The phase drift can be quantified by pattern informatics through calculating the seismic activity over a period of time, change interval, and normalizing over the seismic activity of the whole background. To make a standard process for the PI forecast of M>6 events, we examined the PI pattern before large earthquakes (M>6) that occurred during 2001 to 2010 in Taiwan by giving different cut magnitude from 3.0 to 4.0 and change interval. Cut magnitude is a threshold to keep the catalog complete and the lower limit of the moderate earthquakes. We only use the moderate earthquakes to make PI because the moderate earthquakes can reflect the seismic anomalies most. Our results show that the locations with high anomalies are usually connected with the large earthquakes well when the cut magnitude is 3.2 and the change interval is 4 years. Therefore, the PI method can be utilized as a forecast tool with regularly update, says perform the PI calculation every season.