A Case Study of the Plum Earthquake Early Warning Algorithm Using Southern California Data

Date: 4/25/2019

Time: 06:00 PM

Room: Grand Ballroom

Earthquake early warning (EEW) studies use seismic data to identify large ground shaking generated by earthquakes, with the aim to alert the public. Here, we explore if the Japanese Propagation of Locally Damped Motion (PLUM) EEW algorithm can successfully be applied to Southern California data. Unlike traditional EEW methods, which estimate the earthquake location and magnitude, PLUM simply identifies regions that are likely to experience large ground motions and issues alerts to those regions (Kodera et al., 2016). We apply PLUM to six years of M≥3.5 Southern California events (2012-2017; 193 events), in addition to 49 challenging earthquakes/signals that were used to test the current ShakeAlert algorithms. Instead of using the original Japanese ground motion intensity measure to identify large ground motions, we use the Modified Mercalli Intensity (MMI) scale. We explore computing MMI using various combinations of peak ground acceleration (PGA) and peak ground velocity (PGV), and also test how to optimally incorporate vertical and horizontal component data. Our results favor using both PGA and PGV values, computing the final MMI value from the weighted sum of PGA and PGV that are each computed by combining the three-components of ground motion in quadrature. Using this MMI derivation, we present results from six differently configured instances of the PLUM algorithm, exploring the benefits of using either one or two stations for alert generation, with a variety of MMI alert thresholds. We find that to optimally identify magnitude 5.0+ earthquakes, while suppressing false alerts, two-station methods perform best, favoring a MMI≥4 and MMI≥2.5 alert threshold for the first and second stations, respectively. Our findings also highlight the fact that larger magnitude events do not always generate the largest ground motions.

Presenting Author: Debi Kilb

Authors