Beamforming Out-of-Network Earthquakes on Short DAS Cable Subsegments for Earthquake Early Warning

Theresa M. Sawi, Jeff J. McGuire, Andrew J. Barbour, Clara E. Yoon, & James W. Atterholt

Submitted September 7, 2025, SCEC Contribution #14587, 2025 SCEC Annual Meeting Poster #TBD

Earthquake Early Warning Systems operate by detecting, locating, and estimating the magnitudes of local and regional earthquakes, yet earthquakes outside of the seismic network footprint can often lead to erroneous location and magnitude estimates. Beamforming can aid the location of these events by using an array of seismic instruments to estimate backazimuths and apparent slownesses. In this work, we test a beamforming approach to find the backazimuths and apparent slownesses of earthquakes near the Mendocino Triple Junction with data from four relatively short (~1.7 to 3.5 km) subsegments of a 15-km-long distributed acoustic sensing (DAS) cable near Eureka, California.
Our data are from February 2023 to December 2024, including the M7 December 5th, 2024, Offshore Cape Mendocino earthquake. We explore the practical lower limit of the length of the DAS subsegments that deliver reliable results. We utilize the MUSIC method of beamforming on velocity data that has been converted from strain via a regularized frequency-wavenumber rescaling technique. We test a range of beamforming geometries by varying the length of the DAS subsegment used. We find that the backazimuths and slownesses for earthquakes using DAS beamforming from DAS subsegments with lengths of ~2.5 km are most accurate when compared to locations from the U.S. Geological Survey ComCat earthquake catalog (<10 degrees error for backazimuth and <1.25 s/km error in apparent slowness), unless scattered waves dominate the wavefield. Employing relatively short cable subsegments means that an incoming P-wave need not traverse as many sensors to be identified, thus lowering the observation time needed for event detection and location. Moreover, when designing new arrays, shorter cables are more cost-effective, requiring less expensive interrogators, less cable, trenching, and land. As DAS arrays become more commonplace, understanding the practical limits on the cable geometry needed for beamforming may be a key piece of information when selecting dark fibers or designing new DAS arrays.

Key Words
Earthquake early warning, DAS, array processing

Citation
Sawi, T. M., McGuire, J. J., Barbour, A. J., Yoon, C. E., & Atterholt, J. W. (2025, 09). Beamforming Out-of-Network Earthquakes on Short DAS Cable Subsegments for Earthquake Early Warning. Poster Presentation at 2025 SCEC Annual Meeting.

Related Projects & Working Groups
Seismology