Precursory Seismic Migration Patterns Examined by Improved Pattern Informatics Method

The pattern informatics (PI) method is a statistical method for forecasting earthquakes based on the concept of pattern dynamics. It could detect precursory seismic activation or quiescence. Here we apply the PI method with the frequency- magnitude distribution (FMD) of events, based on the self-organized spinodal (SOS) behavior, on the Chi-Chi and Pingtung earthquakes using the earthquake catalog provided by the Central Weather Bureau. By the FMD test, two important times before a large earthquake when the number of intermediate-sized events has significant increase are found. The PI hotspots which indicate the location with anomalous seismicity are around the epicentral area in stage 2 of the SOS behavior. When it proceeded to stage 3, the hotspots migrate to the epicenter. This migration is not only observed in the PI maps but also quantified and visualized from the calculation of the distances between the epicenter and hotspots. Investigating the evolution of the distance in time yields a slope that indicates the trend suggesting how the hotspots propagate relative to sites. After imaging all slopes on a map, we can get a donut-like pattern that illustrates the migration process. In general, migration occurs intensively in stage 2. The migration process might be introduced by the nucleation of earthquakes, and its duration and range depend on the size of the mainshock. After migration in stage 2, the number of intermediate-sized events increases during stage 3, called the precursory period, in SOS behavior. Such an increase occurs on the epicentral region and might be associated with accelerating seismicity observed in many studies. The improved PI method shows that anomalous precursor signature would be exposed by the seismicity rate not only in a static form but also in a dynamic form. The static form means that the PI hotspots point out a location with high hazard, while the dynamic form shows the preparing process (migration) before a large earthquake. Thus, the improved PI method can be a promising tool for understanding the physics of earthquakes.