← Back to Sessions

Earthquake Detection of the MiDAS Seismic Monitoring System Containing Downhole Optic-Fiber Distributed Acoustic Sensing and Borehole Seismometers

Description: 

Milun fault Drilling and All-inclusive Sensing (MiDAS) project initiated after the destructive 2018 M6.4 Hualien, Taiwan, earthquake. The project successfully drilled through the Milun fault, which was one of the main rupture faults in the Hualien earthquake, in Hole A (700 m deep) on the hanging wall. In addition to this primary hole, a 500-m-deep hole (Hole B) was done on the footwall 700 m away from Hole A. After drilling, the MiDAS group deployed a 7.5-km-length optic-fiber cable perpendicular to the Milun fault, passing through Hole A (from top to bottom), the main fault zone on the surface, and Hole B (from top to bottom) orderly. The downhole triaxial seismometer arrays were installed in both holes. In this study, we search earthquakes from the continuous seismic records by two methods; one is manually picking and another is by the deep-neural-network-based seismic arrival-time picking method, PhaseNet. The study period is from March to May, 2023. We group earthquakes into 5 classifications by the S-P arrival time differences (ts-tp). Class S, A, B, and C demonstrate events with ts-tp ranges < 1s, > 1s & < 2s, > 2s & < 6s, and > 6s, respectively. Class D collects events with unclear P- or S-wave identification. Compared both detected results of human made and from PhaseNet, we recognize that those from PhaseNet have high accuracy (80 ~ 98%) in local events (Class A-C). However, the accuracy drops to 22% for Class S, which includes ultra-close events on the fault. It is possibly due to lack of ultra-close event templates in the PhaseNet model training. Next step, we will improve the PhaseNet model and develop a labor-saving procedure in earthquake detection for the MiDAS seismic monitoring system.

Session: Advancing Seismology with Distributed Fiber Optic Sensing [Poster Session]

Type: Poster

Date: 5/3/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Yen-Yu

Student Presenter: No

Invited Presentation: 

Authors