Event Detection Performance of the PLUM Earthquake Early Warning Algorithm in Southern California

Abstract

We test the Japanese ground‐motion‐based earthquake early warning (EEW) algorithm, propagation of local undamped motion (PLUM), in southern California with application to the U.S. ShakeAlert system. In late 2018, ShakeAlert began limited public alerting in Los Angeles to areas of expected modified Mercalli intensity (IMMI) 4.0+ for magnitude 5.0+ earthquakes. Most EEW systems, including ShakeAlert, use source‐based methods: they estimate the location, magnitude, and origin time of an earthquake from P waves and use a ground‐motion prediction equation to identify regions of expected strong shaking. The PLUM algorithm uses observed ground motions directly to define alert areas and was developed to address deficiencies in the Japan Meteorological Agency source‐based EEW system during the 2011 Mw 9.0 Tohoku earthquake sequence. We assess PLUM using (a) a dataset of 193 magnitude 3.5+ earthquakes that occurred in southern California between 2012 and 2017 and (b) the ShakeAlert testing and certification suite of 49 earthquakes and other seismic signals. The latter suite includes events that challenge the current ShakeAlert algorithms. We provide a first‐order performance assessment using event‐based metrics similar to those used by ShakeAlert. We find that PLUM can be configured to successfully issue alerts using IMMI trigger thresholds that are lower than those implemented in Japan. Using two stations, a trigger threshold of IMMI 4.0 for the first station and a threshold of IMMI 2.5 for the second station PLUM successfully detect 12 of 13 magnitude 5.0+ earthquakes and issue no false alerts. PLUM alert latencies were similar to and in some cases faster than source‐based algorithms, reducing area that receives no warning near the source that generally have the highest ground motions. PLUM is a simple, independent seismic method that may complement existing source‐based algorithms in EEW systems, including the ShakeAlert system, even when alerting to light (IMMI 4.0) or higher ground‐motion levels.

The PLUM Earthquake Early Warning Algorithm: A Retrospective Case Study of West Coast, USA, Data

“…Our goal is detecting M5+ earthquakes, where timely detections are more likely to be possible, so we favor Instance #5 that detects the larger magnitude events while reducing the detection of smaller events. This is the main driver in selecting Instance #5 over Instance #6 and explains why, counter‐intuitively, when examining data from the full West Coast that has lower station density than the prior southern California study (Cochran et al., 2019; Instance #6 preferred) that we prefer an increase, rather than a decrease, of the second triggering threshold.…”

“…Cochran et al. (2019) found that using two stations with primary and secondary triggering thresholds of I MMI 4.0 and 2.5, respectively, was preferred to reduce false detections and maximize event detection for southern California data. In this work, the focus is on PLUM event detections and PLUM detection times, computation and evaluation of PLUM alert regions is not performed, as this is a secondary task left for future studies.…”

“…Here, we evaluated the same six instance configurations as in Cochran et al. (2019), which tested southern California data, to determine which configuration is optimal for the west coast of the United States.…”

“…Following Cochran et al. (2019), the first criteria is based on timing, where we require the first station trigger to occur after the arrival of the P‐wave (with assumed speed of 6.5 km/s) and no later than the passing of surface waves (assumed speed 2 km/s). To account for variations in seismic velocities throughout our study region, and the one‐second ambiguity in the time of the trigger, we assume the P‐wave might arrive up to one second earlier than expected and that the surface wave might arrive up to 5 s later than expected.…”

“…Here, we use a modified version of the PLUM algorithm first tested with Southern California Seismic Network data data by Cochran et al. (2019). We have improved the algorithm to select more reliable data at co‐located sensors and have reevaluated triggering thresholds to reliably detect M5+ earthquakes while minimizing detections of smaller events for which timely detections are often not possible (Kodera, 2019).…”