Near-Surface Characterization Using Localized Seismic Signals Acquired by Fiber-Optic Distributed Acoustic Sensing

Abstract Distributed acoustic sensing (DAS) provides unprecedentedly dense wavefield observations for high-resolution near-surface structural imaging, yet its application often relies on specific active sources or ideal noise conditions. To address this limitation, this study proposes a workflow that leverages localized seismic signals, with human footsteps used as a representative source type. Our approach integrates an object detection algorithm to automatically identify and extract footstep signals from continuous DAS recordings. We then construct datasets from numerous events, extract and stack the dispersion spectra of subarrays using multichannel analysis of surface waves, and obtain fundamental-mode surface-wave dispersion curves. Finally, the near-surface velocity structure underneath the study area is obtained. The imaging results reveal distinct velocity variations along different cable segments that are broadly consistent with outcropped geological conditions, supporting the feasibility of the workflow. This approach provides a cost-effective and environmentally adaptive imaging solution that operates without relying on conventional sources. Although demonstrated using footstep signals, the workflow may be adapted to other suitable localized anthropogenic or controlled sources, potentially expanding DAS applicability in complex settings.