Abstract
AbstractSubduction zones are monitored using space geodesy with increasing resolution, with the aim of better capturing the deformation accompanying the seismic cycle. Here, we investigate data characteristics that maximize the performance of a machine learning binary classifier predicting slip‐event imminence. We overcome the scarcity of recorded instances from real subduction zones using data from a seismotectonic analog model monitored with a spatially dense, continuously recording onshore geodetic network. We show that a 70–85 km‐wide coastal swath recording interseismic deformation gives the most important information on slip imminence. Prediction performances are mainly influenced by the alarm duration (amount of time that we consider an event as imminent), with density of stations and record length playing a secondary role. The techniques developed in this study are most likely applicable in regions of slow earthquakes, where stick‐slip‐like failures occur at time intervals of months to years.
Highlights
The preparatory phase of large subduction earthquakes can be depicted as a period of slow, continuous stress accumulation caused by the frictional interaction between converging plates (e.g., Hyndman et al, 1997)
We investigated the role of space‐time coverage and alarm duration on the performance of analog earthquake prediction
We found that alarm duration plays a primary role in tuning the performances of a binary classifier
Summary
The preparatory phase of large subduction earthquakes can be depicted as a period of slow, continuous stress accumulation caused by the frictional interaction between converging plates (e.g., Hyndman et al, 1997). As geodetic (Global Navigation Satellite System [GNSS]) observation networks have matured, it has become apparent that there are significant variations of interplate locking before and/or after large earthquakes These include transient slow slip events, indicating sub seismic‐cycle scale, short (days to months) variations in the rates of stress accumulation/release (Heki & Mitsui, 2013; Loveless & Meade, 2016; Mavrommatis et al, 2014; Melnick et al, 2017). Stress variations prior to large earthquakes may manifest as a series of foreshocks gradually unzipping the plate interface—as in the case of the 2014 Iquique M 8.1 earthquake (Schurr et al, 2014)—or as accelerating aseismic creep—as suggested for the 2011 Tohoku M 9.0 earthquake (Kato et al, 2012; Mavrommatis et al, 2014) The recognition of these pre‐earthquake transients raises the potential for using them as a diagnostic tool for earthquake imminence. The scarcity of recorded instances hinders understanding whether and which transient signal may be used as a reliable indicator for earthquake prediction
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have