Analysis of saturation effects of distributed acoustic sensing and detection on signal clipping for strong motions

SUMMARY Distributed acoustic sensing (DAS) systems are increasingly used for earthquake monitoring due to their cost-effectiveness and high spatial resolution. However, signals exceeding the dynamic range in DAS systems lead to signal clipping and data loss during strong ground motion and near-fault observations. In this study, we investigated the saturation effects of DAS signal clipping using two collocated DAS arrays with a looped setup in Hualien City, drawing on seismic data from the 2022 $M_{\rm W}$ 7.06 Taitung earthquake sequence. The two DAS arrays, connected to different interrogators, simultaneously recorded the earthquake signals and exhibited different dynamic ranges, allowing for direct comparisons of clipped and unclipped signals. Our results indicate that the primary factors contributing to signal clipping in DAS can be categorized as (1) strong ground motion induced by earthquake magnitude and cable installations and (2) the limited dynamic range of the interrogator. Furthermore, our analysis reveals that signal clipping leads to an amplitude increase across all frequencies in the spectra, resembling the addition of a white-noise-like signal that contaminates the waveform spectra. To address this issue, we develop a frequency-based detection approach using spectral coherence estimation on collocated channels to identify clipped signals. Our findings demonstrate that coherencegrams can be employed to detect clipped signals to ensure the reliability of DAS data during strong ground motion and enhance applications that rely on near-real-time high-quality data, such as earthquake early warning systems.