Building New Earthquake Catalogs Using Machine Learning-based Seismic Phase Picking Algorithms in Northern California
Abstract
Machine learning-based seismic phase picking algorithms are well-developed and widely-accepted new methods in the seismology community. They have been shown to have better performance than traditional phase picking methods in many ways. For example, the ability to pick seismic phases with lower signal-to-noise levels lowers the earthquake detection threshold allowing us to build a more complete earthquake catalog, which can aid in the study of earthquake cycles and activate tectonics. Researchers have been using these algorithms in many case studies including specific fault zones or earthquake sequences. We applied different machine learning-based seismic phase picking algorithms (PhaseNet (Zhu and Beroza, 2019), Earthquake transformer (Mousavi et al., 2020), Generalized Seismic Phase Detection (Ross et al, 2018)) to the whole Northern California seismic network. With one years data from 2020, we are able to test and compare the performances of these different machine learning algorithms. Then we used the same phase association and locating methods as the current Northern California network to build a new earthquake catalog. We tried different ways to test the reliability of the machine learning algorithms. In some special areas where we had earthquake clusters, we compared the machine learning algorithms to template matching methods. We also integrated results from different machine learning algorithms to improve the reliability of detected earthquakes. In summary, we will present a quantitative comparison of different machine learning-based seismic phase picking algorithms on the Northern California dataset and the new earthquake catalogs they generate. The results provide a more complete understanding of this regions seismicity and provide insights on the use of these machine learning algorithms.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.S32A..01G