Deep Learning for Deep Earthquakes: Insights from OBS Observations of the Tonga Subduction Zone

This is a Preprint and has not been peer reviewed. The published version of this Preprint is available: https://doi.org/10.1093/gji/ggae200. This is version 1 of this Preprint.

Add a Comment

You must log in to post a comment.


Comments

There are no comments or no comments have been made public for this article.

Downloads

Download Preprint

Authors

Ziyi Xi, Songqiao Shawn Wei, Weiqiang Zhu, Greg Beroza, Yaqi Jie, Nooshin Saloor

Abstract

Applications of machine learning in seismology have greatly improved our capability of detecting earthquakes in large seismic data archives. Most of these efforts have been focused on continental shallow earthquakes, but here we introduce an integrated deep-learning-based workflow to detect deep earthquakes recorded by a temporary array of ocean-bottom seismographs (OBSs) and land-based stations in the Tonga subduction zone. We develop a new phase picker, PhaseNet-TF, to detect and pick P- and S-wave arrivals in the time-frequency domain. The frequency-domain information is critical for analyzing OBS data, particularly the horizontal components, because they are contaminated by signals of ocean-bottom currents and other noise sources in certain frequency bands. PhaseNet-TF shows a much better performance in picking S waves compared to its predecessor PhaseNet. The predicted phases are associated using an improved Gaussian Mixture Model Associator GaMMA-1D and then relocated with a double-difference package teletomoDD. We further enhance the model performance with a semi-supervised learning approach by iteratively refining labelled data and retraining PhaseNet-TF. This approach effectively suppresses false picks and significantly improves the detection of small earthquakes. The new catalogue of Tonga deep earthquakes contains more than 10 times more events compared to the reference catalogue that was analyzed manually. This deep-learning-enhanced catalogue reveals Tonga seismicity in unprecedented detail, and better defines the lateral extent of the double-seismic zone at intermediate depths and the location of 4 large deep-focus earthquakes relative to background seismicity. It also offers new potential for deciphering deep earthquake mechanisms, refining tomographic models, and understanding of subduction processes.

DOI

https://doi.org/10.31223/X5C105

Subjects

Earth Sciences, Geophysics and Seismology, Physical Sciences and Mathematics

Keywords

machine learning, seismicity and tectonics, Subduction zone processes, Neural Networks, Pacific Ocean

Dates

Published: 2023-11-06 04:47

Last Updated: 2023-11-06 09:47

License

CC-By Attribution-NonCommercial-NoDerivatives 4.0 International

Additional Metadata

Data Availability (Reason not available):
All data and codes associated with this study are publicly accessible and built upon open-source platforms. PhaseNet-TF can be accessed via its GitHub repository: https://github.com/ziyixi/PhaseNet-TF, and GaMMA-1D is available at https://github.com/AI4EPS/GaMMA. PhaseNet and GaMMA packages are available at their respective GitHub repositories: https://github.com/AI4EPS/PhaseNet and https://github.com/AI4EPS/GaMMA (the same as GaMMA-1D). More specifically, the PhaseNet-TF model weights for Tonga deep earthquakes are available at https://github.com/ziyixi/PhaseNet-TF/releases/tag/v0.3.0. Data processing and visualization are conducted using open-source Python packages, including ObsPy (https://github.com/obspy/obspy/) and PyGMT (https://github.com/GenericMappingTools/pygmt). Seismic data in this research are archived at the EarthScope/IRIS Data Management Center under network codes YL, II, and Z1.