Exploring the Limitations and Solutions for Large Earthquake Observations with DAS

Distributed Acoustic Sensing (DAS) has become a widely used tool for measuring the strain of seismic wavefields with high spatial and temporal resolution along a long-distance fiber-optic cable. However, there is still a lack of knowledge in using DAS data for large earthquake observations, such as strain-velocity conversion and instrumental recording limitations. A common issue in large earthquake observations is the saturation of seismic waveforms, which exceed the dynamic range of instruments. This saturation effect causes signal clipping and difficulties in determining earthquake magnitude for earthquake early warning, performing advanced spectral analysis, and assessing ground-motion variability and associated uncertainties. In this study, we deployed two collocated DAS interrogators connected to fiber-optic cable with different cable installation styles and a broadband seismometer to record ground motion from an earthquake sequence located 100 km away, with magnitudes ranging from M_W 3.81 to 7.06. We demonstrate that signal clipping in DAS is influenced not only by earthquake magnitude but also by interrogator type and cable coupling. We observe that signal clipping in DAS records increases amplitudes at all frequencies in the signal spectra, resulting in white noise-like effects. We develop a frequency-based approach using multitaper spectral analysis and time-varying coherence estimation between interrogators to identify signal clipping instances within the DAS dataset and ensure data quality control in large earthquake observations. With our understanding of the saturation effect in records, we can devise strategies and solutions based on gauge length, sampling rate, and local phase velocity to avoid signal clipping when using DAS for large earthquake observations.