← Back to Gallery

Detection of Atypical Events in Continuous Downhole Distributed Acoustic Sensing (DAS) Monitoring Using Deep Learning

Distributed Acoustic Sensing (DAS) is an emerging technology that converts standard optical fiber into dense array of virtual sensors, enabling high‐resolution detection of acoustic signals along the entire length of optical fiber. An M_W 6.4 earthquake that struck eastern Taiwan on 6 February 2018 highlighted the seismic hazard posed by the Milun Fault, as demonstrated by severe structural damage in Hualien City. To investigate fault structures and seismic activity in this region, the Milun Fault Drilling and All‐inclusive Sensing (MiDAS) project drilled a borehole across the fault zone. A downhole fiber‐optic cable was deployed and connected to a DAS interrogator for continuous monitoring. The system recorded both earthquakes and a series of atypical signals. These atypical signals distinguished by short durations and rapid attenuation that render them challenging to identify using conventional approaches. To address this challenge and investigate the origin of atypical events, we developed a deep learning framework and applied it to three months of continuous downhole DAS data. The framework successfully detected both earthquakes and atypical signals, with the latter concentrated between 250-360 m depth and cautiously attributed to transient disturbances arising from the early-stage hydration of lower-density cement grout. Their frequency declined markedly following cement solidification, representing the first documented DAS observations of signals linked to this process. More broadly, our findings highlight the potential of integrating deep learning with DAS to detect transient vibration signals, including those generated by fracture processes during gas or fluid injection, underscoring its value for advancing subsurface monitoring and hazard assessment.

Session: Fiber-Optic Sensing Applications in Seismology and Environmental Science [Poster]

Type: Poster

Room: Exhibit Hall A+B

Date: 4/17/2026

Presentation Time: 08:00 AM (local time)

Presenting Author: Chin-Shang Ku

Student Presenter: No

Invited Presentation: 

Poster Number: 77

Additional Authors