Detection of Cascadia Slow Slip in Borehole Pore Pressure Data

Abstract

ABSTRACT The Cascadia subduction zone hosts quasi-periodic, large-scale slow slip events which have been shown to be precursors to large, tsunamigenic earthquakes at other active subduction zone margins. Detection of these events in Cascadia utilizes continuous GPS measurements and high-resolution seismic stations to capture geodetic reversals and spatially correlated tremor, respectively, associated with slow slip. Borehole pore pressure data have been shown to record signals related to slow slip and may provide an alternate means of detection for these events, potentially in near real-time. We evaluate this hypothesis by developing and testing anomalous signal detection methods for these data, specifically using evaluations of STA/LTA and a Holt-Winter’s predictive model. We also perform Bayesian statistical analysis on these detections to determine the posterior probability of a slow slip event given an anomalous pore pressure detection. Our results show that a Holt-Winter’s model readily detects strongly correlated anomalous activity across multiple boreholes at known times of slow-slip, and the posterior probability of slow slip given a detection ranges from 50-100%. We conclude that borehole pore pressure data do record anomalous signals associated with slow slip, and we speculate that groups of multiple, closely located boreholes may perform as a real-time detection network for slow slip events.