On the Statistical Significance of Foreshock Sequences in Southern California

M P A Van Den Ende,J.‐P Ampuero

doi:10.1029/2019gl086224

Abstract

AbstractEarthquake foreshocks may provide information that is critical to short‐term earthquake forecasting. However, foreshocks are far from ubiquitously observed, which makes the interpretation of ongoing seismic sequences problematic. Based on a statistical analysis, Trugman and Ross (2019, https://doi.org/10.1029/2019GL083725) suggested that as much as 72% of all mainshocks in Southern California is preceded by foreshock sequences. In this study, we reassess the analysis of Trugman and Ross (2019, https://doi.org/10.1029/2019GL083725), and we evaluate the impact of the assumptions made by these authors. Using an alternative statistical approach, we find that only 15 out of 46 mainshocks (33%) are preceded by significantly elevated seismicity rates. When accounting for temporal fluctuations in the background seismicity, only 18% of the analyzed foreshock sequences remain unexplained by the background seismicity. These results imply that even in a highly complete earthquake catalog, the majority of earthquakes do not exhibit detectable foreshock sequences.

Highlights

Prior to large earthquakes, precursory signals, such as accelerating aseismic creep and elevated tremor and foreshock activity (Dodge et al, 1996; Jones & Molnar, 1979; Marsan & Enescu, 2012), may be recorded by GPS and seismic stations
While it has been argued that some, if not most, large earthquakes are preceded by detectable foreshock sequences (Bouchon et al, 2013), elevated seismicity rates do not uniquely signify the advent of an earthquake, as many of these excursions return to the background seismicity rate
The reported results do not immediately illuminate the significance of elevated seismicity rates when taking into perspective the ubiquitous fluctuations in the background seismicity rate: if similar elevations in the seismicity rate are observed at random 70% of the time, the presence of a foreshock sequence may be due to random chance

Summary

Introduction

Precursory signals, such as accelerating aseismic creep (see Roeloffs, 2006) and elevated tremor and foreshock activity (Dodge et al, 1996; Jones & Molnar, 1979; Marsan & Enescu, 2012), may be recorded by GPS and seismic stations. While it has been argued that some, if not most, large earthquakes are preceded by detectable foreshock sequences (Bouchon et al, 2013), (locally) elevated seismicity rates do not uniquely signify the advent of an earthquake, as many of these excursions return to the (average) background seismicity rate These observations prohibit the utilization of seismicity rates in earthquake forecasting attempts to date. By comparing short-term seismicity rates with the background rate over 1 year prior to a selected mainshock, T&R concluded that over 70% of all analyzed mainshocks were preceded by a statistically significant increase in seismicity rates These authors further alluded to the possibility that in practically all cases foreshock sequences may be detected, provided that the earthquake catalog is sufficiently complete. The reported results do not immediately illuminate the significance of elevated seismicity rates when taking into perspective the ubiquitous fluctuations in the background seismicity rate: if similar elevations in the seismicity rate are observed at random 70% of the time, the presence of a foreshock sequence may be due to random chance

Methods

Results

Discussion

Conclusion