Improving Detections by Reducing Problematic Triggers in the Epic Earthquake Early Warning Algorithm
Description:
One of the greatest challenges of developing an Earthquake Early Warning (EEW) system such as the US West Coast’s ShakeAlert system is the ability to know how the system will perform during a large, never-before-seen earthquake. While historical records exist, like those that comprise the ShakeAlert Testing and Certification Testsuite (Cochran et al., 2018), they are primarily from older, out-of-network, and/or international earthquakes. As the network of stations used by ShakeAlert differs from that which recorded many of these earthquakes (e.g. increased density of stations, stations added/removed/upgraded, etc.), running replays of these earthquakes through the ShakeAlert system does not accurately reflect the performance of the current ShakeAlert system. Furthermore, the system can only be tested using data from earthquakes that have already occurred. For example, it is not possible to know how ShakeAlert will perform on a large earthquake on the Hayward Fault with our current seismic network.
The U.S. Department of Energy EarthQuake SIMulation (EQSIM) is a novel simulation framework capable of creating regional fault-to-structure simulations of earthquakes with unprecedented fidelity. These simulations are computed on the DOE’s GPU-accelerated exaflop platforms and can contain hundreds of billions of model grid points. With grid point spacing of 6.25 m and 5-Hz resolution, these simulations are a unique dataset that can be used to test the ShakeAlert system with its current network configuration without having to wait for a large, damaging earthquake to occur.
We tested the feasibility of using this dataset to test EEW algorithms by replaying five unique EQSIM simulations of a M7 rupture on the Hayward fault with the ShakeAlert EEW point-source algorithm EPIC. Here, we present the results of those replays and explore the potential of using such high-fidelity simulations for EEW algorithm testing and development.
Session: Performance and Progress of Earthquake Early Warning Systems Around the World - II
Type: Oral
Date: 4/16/2025
Presentation Time: 11:30 AM (local time)
Presenting Author: Angie
Student Presenter: No
Invited Presentation:
Poster Number:
Authors
Angie Lux Presenting Author Corresponding Author angiechung07@gmail.com University of California, Berkeley |
Ivan Henson ihenson@berkeley.edu University of California, Berkeley |
Andrei Akimov andrei.akimov@berkeley.edu University of California, Berkeley |
Richard Allen rallen@berkeley.edu University of California, Berkeley |
David Mccallen dbmccallen@lbl.gov Lawrence Berkeley National Laboratory |
Arben Pitarka pitarka1@llnl.gov Lawrence Livermore National Laboratory |
Houjun Tang htang4@lbl.gov Lawrence Berkeley National Laboratory |
|
|
Improving Detections by Reducing Problematic Triggers in the Epic Earthquake Early Warning Algorithm
Category
Performance and Progress of Earthquake Early Warning Systems Around the World