STATISTICAL CHARACTERISTICS FOR THE BACKGROUND NOISE IN DISTRIBUTED ACOUSTIC SENSING: ANALYSIS AND APPLICATION TO SUPPRESSION

T. ZHONG,Y. CHEN,X. T. DONG,Y. LI,N. WU

2022-01-01

Journal of seismic exploration

Abstract:Distributed acoustic sensing (DAS) is a novel technology that utilizes a fiber-optic cable instead of geophones, which has attracted increasing attention in seismic data acquisition. However, owing to the existence of background noise, the current quality of the DAS records requires improvement. In this study, the stationarity and spectral characteristics for DAS background noise are investigated. Additionally, the dataset used for the analysis is collected while satisfying the practical requirements of the exploration industry. The results demonstrate that the DAS background noise is a broadband interference with local stationarity. On this basis, an adaptive time-frequency peak filtering (TFPF) algorithm is proposed to attenuate the background noise. Unlike traditional TFPF algorithms, this improved method adaptively chooses appropriate filtering parameters instead of using a fixed parameter set to the whole seismic record to achieve better attenuation performance. Specifically, the signal and noise segments can be recognized by taking advantage of the differences in stationarity. Consequently, we can adaptively select different filtering parameters for signal and noise segments to get better performance in noise attenuation and signal restoration. Synthetic and field data experimental results indicate that the proposed adaptive TFPF algorithm can suppress the DAS background noise and accurately recover the reflection events, especially under low signal-to-noise ratio conditions.