Testing the Pattern Informatics index on synthetic seismicity catalogs based on the Non-Critical PAST

Abstract

Listen

Translate article

The Non-Critical Precursory Accelerating Seismicity Theory (or Non-Critical PAST) has recently been proposed to explain the formation of accelerating seismicity that may be observed before large earthquakes. It has led to the possibility of constructing synthetic seismicity catalogs where patterns of accelerating seismicity (∼ activation) and quiescence, which occur in the same space–time window, can be evaluated by existing forecasting techniques. In this study, the Pattern Informatics (or PI) index is tested on synthetic catalogs where a realistic spatiotemporal clustering has been added on top of the theoretical precursory seismicity. We show that the PI index is successful at identifying the precursory quiescent signal but fails in identifying precursory accelerating seismicity directly, being more sensitive to aftershock sequences of background events than to the activation-like behavior of the acceleration, formed by background events alone. We also show that the PI index has a high success rate in finding precursory quiescence, even for a low signal-to-noise ratio. As for the few false negatives, they are usually due to interfering aftershocks as well, which skew seismicity rates to higher means. The Non-Critical PAST, by helping to better understand the behavior of the PI algorithm in synthetic catalogs, gives new perspectives on how to improve it and on how to use it in real catalogs.