14:00 〜 14:15
[SSS06-02] Noise Attenuation in Distributed Fiber-Optic Sensing Data Using a Spectral Subtraction-based Approach
キーワード：Distributed Fiber-Optic Sensing, Spectral Subtraction, Noise Reduction
Distributed Fiber-Optic Sensing (DFOS) technology has emerged as an effective data acquisition tool for seismological applications, particularly in microseismic monitoring. Its unique capability to convert fiber-optic cables into dense seismic arrays offers numerous advantages over conventional seismic networks, especially in challenging environments such as deep boreholes in Carbon Capture and Storage (CCS) or Enhanced Geothermal Systems (EGS) projects.
However, seismic signals recorded by DFOS systems are usually characterized by higher noise levels than those recorded by standard seismic sensors (e.g., geophones). A key limitation of working with DFOS data is that traditional filtering methods often fail to recover weak signals, leading to suboptimal noise reduction performance. This work introduces a robust denoising approach that adapts speech signal processing techniques based on spectral subtraction to enhance DFOS data quality.
We validate this method using synthetic DFOS data simulating realistic data acquisition geometries and noise conditions. The denoising workflow is then applied to real microseismic DFOS data recorded during the April 2022 stimulation campaign at the FORGE EGS site in Utah, USA.
Our results, show significant improvements in Signal-to-Noise Ratio (SNR), demonstrating the effectiveness of our method even in low-SNR conditions. This workflow outperforms traditional filtering techniques, offering a promising solution for enhancing DFOS data quality and improving the detection of previously hidden signals.
However, seismic signals recorded by DFOS systems are usually characterized by higher noise levels than those recorded by standard seismic sensors (e.g., geophones). A key limitation of working with DFOS data is that traditional filtering methods often fail to recover weak signals, leading to suboptimal noise reduction performance. This work introduces a robust denoising approach that adapts speech signal processing techniques based on spectral subtraction to enhance DFOS data quality.
We validate this method using synthetic DFOS data simulating realistic data acquisition geometries and noise conditions. The denoising workflow is then applied to real microseismic DFOS data recorded during the April 2022 stimulation campaign at the FORGE EGS site in Utah, USA.
Our results, show significant improvements in Signal-to-Noise Ratio (SNR), demonstrating the effectiveness of our method even in low-SNR conditions. This workflow outperforms traditional filtering techniques, offering a promising solution for enhancing DFOS data quality and improving the detection of previously hidden signals.