Deep learning-enhanced catalog of microseismicity in The Geysers geothermal field

Lei Li, Ian W. McBrearty, Xing Tan, William L. Ellsworth, & Gregory C. Beroza

Published September 8, 2024, SCEC Contribution #13753, 2024 SCEC Annual Meeting Poster #028 (PDF)

Induced seismicity monitoring with local and dense arrays has shown great potential for fine-scale fracture/fault characterization, constraining mechanisms of earthquake triggering, and for seismic risk management. The Geysers geothermal field in northern California is the largest geothermal field worldwide in terms of steam production. The operations there have induced or triggered several tens of thousands of microseismic events every year. In this study, we characterize induced microseismicity during the past three years (2021-2023) with deep learning-based workflows (e.g., LOC-FLOW) applied to continuous waveforms from the local network BG (consisting of 37 three-component stations) to automatically detect and locate microseismic events. We refine the locations using HypoDD and calculate local magnitudes (ML) of these events based on a local magnitude estimation formula suitable for northern and central California. We have obtained more than three times more microseismic events (~60,000 events per year) compared to the original NCEDC catalog (~16,000 events per year), with a magnitude of completeness around 0.2. The deep learning-enhanced catalog enables an improved microseismic analysis, including event clusters, b-value monitoring, and correlations between microseismic activity and geothermal operations.

Key Words
deep learning, induced seismicity, microseismicity, Geysers

Citation
Li, L., McBrearty, I. W., Tan, X., Ellsworth, W. L., & Beroza, G. C. (2024, 09). Deep learning-enhanced catalog of microseismicity in The Geysers geothermal field. Poster Presentation at 2024 SCEC Annual Meeting.

Related Projects & Working Groups
Seismology