← Back to Sessions

Real-time GNSS Data in ShakeAlert: Potential Improvements for Subduction Megathrust Earthquakes Through Network Design and Distributed Slip Models

Description: 

The ShakeAlert® earthquake early warning system characterizes an earthquake’s location, magnitude, and rupture extent with three algorithms and calculates anticipated ground shaking from these source parameters. The EPIC algorithm uses the first few seconds of the P-wave to estimate magnitude but can exhibit magnitude saturation for M7+ events. The FinDer algorithm applies template matching to observed ground acceleration to infer the location, orientation, and length of a line source. The GFAST-PGD algorithm, in use since March 2024, uses peak ground displacement (PGD) measured by real-time Global Navigation Satellite System (GNSS) data to estimate magnitude, assuming the point source location estimated by ShakeAlert’s seismic algorithms. GFAST-PGD’s inclusion was motivated by its ability to estimate non-saturating magnitudes for very large (e.g., M8+) earthquakes.

I explore the impact of GNSS station distribution on the warning times GFAST-PGD might provide and the potential of distributed slip models derived from real-time GNSS data (rtGNSS DSMs) to improve warning times, both in the context of subduction megathrust earthquakes. For the former I develop a method to rank existing or potential GNSS station locations based on their ability to improve warning times at population centers where earthquakes of a given magnitude on the slab interface are expected to produce shaking of a chosen intensity. The method accounts for the minimum number of stations required by GFAST-PGD and additional criteria for observed PGD that ShakeAlert uses to mitigate the effect of noisy data. Separately, I explore the impact of rtGNSS DSMs obtained with the BEFORES algorithm in a simulated ShakeAlert implementation. Ground motion calculated from the time-evolving slip models can enable longer warning times for locations experiencing MMI 6+ shaking. However, depending on the MMI threshold used to determine which locations are alerted, large geographic regions with only weak shaking would likely receive alerts as well. Performance might be improved by reducing BEFORES’ assumed subfault size and the depth to which rupture may extend.

Session: Performance and Progress of Earthquake Early Warning Systems Around the World [Poster]

Type: Poster

Date: 4/16/2025

Presentation Time: 08:00 AM (local time)

Presenting Author: Jessica

Student Presenter: No

Invited Presentation: 

Poster Number: 121

Authors