← Back to Sessions

Improving Shear-Arrival Time Estimates for Real-Time Association and Location Algorithms

Description: 

Machine learning and, in particular, deep-learning frameworks have markedly changed the capability of regional seismic networks to detect, classify, and quantify seismic phase arrivals. Yet, in real-time seismic monitoring frameworks, our association and location tools are not necessarily well-equipped to utilize these new and improved automatic characterizations. This limitation is particularly true for shear-wave arrivals which deep-learning detectors appear to generate at a rate of approximately equal to or greater than primary-wave arrivals. The main challenge posed by shear-wave arrivals is that in monitoring regions with substantial three-dimensional velocity heterogeneities, predicting shear-arrival times requires a time-intensive tomography process. To mitigate this tomography requirement, University of Utah Seismograph Stations (UUSS) has recently begun to accommodate 3D structural effects with source-specific station corrections (SSSCs). In this abstract, we review ongoing work at UUSS that combines conventional phase-specific station corrections with SSSCs to better predict phase arrivals at regional monitoring scales. The correction estimation and implementation strategies to be presented appear to be effective, safe, conceptually simple, data-driven, low-overhead, and well-suited for integration into typical earthquake association and location frameworks. Time permitting, we will also review the performance of these correction strategies in a live-acquisition, real-time environment.

Session: Network Seismology: Recent Developments, Challenges and Lessons Learned - V

Type: Oral

Date: 5/2/2024

Presentation Time: 02:15 PM (local time)

Presenting Author: Ben

Student Presenter: No

Invited Presentation: 

Authors