The PLUM Earthquake Early Warning Algorithm: Application to Real-Time, USA, Data

Session: Earthquake Early Warning: Current Status and Latest Innovations
Type: Oral
Date: 4/29/2020
Time: 11:00 AM
Room: 115
Description: 

The Propagation of Local Undamped Motion (PLUM) earthquake early warning (EEW) algorithm, originally designed for use in Japan, is also effective at detecting quakes in the Western United States. Here, we apply PLUM to real-time data from California, Oregon and Washington, USA, exploring if PLUM could contribute to the ShakeAlert EEW system. Differing from current ShakeAlert algorithms, PLUM is ground-motion-based and designed to issue alerts when the observed ground motions exceed a specified intensity. PLUM only reports the location of large ground motions; it does not report earthquake magnitude or location as in the current ShakeAlert algorithms. In 2018, a real-time version of PLUM was implemented on an internal ShakeAlert development server, which allowed us to improve methods of identifying and mitigating data streaming gaps and bad data. Currently, PLUM requires two neighboring stations to observe ground motions exceeding MMI 4.0 and MMI 2.5 for the first and second stations, respectively. PLUM alerted on the 2019 July Ridgecrest M6.4 and M7.1 earthquakes, with latencies (including telemetry) of 5 and 6 seconds, on par or better than the production ShakeAlert algorithms EPIC and FinDer. To date, PLUM detected 127 earthquakes and 29 non-earthquake signals. Currently, real-time PLUM is running on two ShakeAlert development servers to assess performance metrics. Consistent with the ShakeAlert ground motion product, we are now updating PLUM to report predicted ground motions on a 0.2°x0.2° grid. We conclude that in regions of dense station coverage (inter-station spacing <30 km), PLUM provides timely information, including flagging relatively small magnitude quakes (M<4) that produce ground motions above the ShakeAlert alerting threshold (MMI>=4).

Presenting Author: Debi Kilb

Authors