Considerations for Modernizing Earthquake Data Analysis at the Southern California Seismic Network
Abstract
Earthquake monitoring systems commonly in use at seismic networks for processing data have been well vetted and established throughout the years. While the current systems continue to perform well, recent advances in computing technologies present opportunities for improving the accuracy and efficiency of data processing.
The Southern California Seismic Network (SCSN) transitioned to the Earthworm-based AQMS software around 2001 and continues to use this system. Earthquakes are automatically picked with an STA/LTA algorithm and located with hypoinverse using a 1-D velocity model. Analysts manually review every detected event and may choose to accept the automatic solution (around 10% of events) or adjust the picks as they deem necessary. Over the past several years, the SCSN catalog has been complete down to M1.8, though it includes many smaller events as low as ~M0. To improve network operations, the SCSN is investigating and developing new data processing systems that take advantage of modern technologies. Machine learning methods can automatically pick phases with a similar accuracy to analysts, thereby holding promise for reducing analyst workload, especially during active earthquake sequences. These methods also perform well at detecting lower magnitude events, allowing for more complete catalogs. Cloud computing provides opportunities for increasing available computational resources and for making systems more robust against failure of local hardware systems or utilities. However, seismic networks must also consider the financial and labor costs of setting up, testing, and maintaining these systems, and ensure catalog continuity and reliable performance with large earthquakes, for which machine learning methods currently identify phases with lower accuracy. We describe some of the SCSN's modernization efforts, including the quakes2aws project that aims to build a cloud-based, modular earthquake analysis system that leverages machine learning methods. We also assess how modernizing the automatic systems could affect the network's workflow, costs, and earthquake catalog product. We hope these considerations may help other networks decide how to move forward with improving their data analysis.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.S25C..06T