Earthquake Early Warning (EEW) systems rely on dense seismic networks to rapidly characterize events, yet their effectiveness is often limited in offshore regions where instrumentation is sparse. Distributed Acoustic Sensing (DAS) offers a novel capability by converting existing subsea fiber-optic cables into dense seismic arrays. We present dEPIC (DAS-Earthquake Point-source Integrated Code), the first operational framework designed to integrate DAS arrays into EEW systems.
Deployed on a submarine cable in Monterey Bay, California, dEPIC utilizes an edge-computing architecture to process high-volume strain rate data in real time. The system employs a GPU-accelerated, machine-learning-based phase picker adapted for DAS, coupled with a grid-search location algorithm and empirical magnitude estimator. To address the azimuthal ambiguity inherent in linear cable geometries, we developed real-time location quality metrics—prominence and distribution scores—that effectively suppress unstable solutions.
Performance evaluations using archived event replays and continuous real-time data demonstrate that dEPIC successfully detects both onshore and offshore events with sub-second processing times. The system operates both autonomously and in coordination with ShakeAlert’s EPIC algorithm, improving detection speed for offshore seismicity. Crucially, since the Israeli “Truaa” EEW system employs this same EPIC algorithm, dEPIC offers a straightforward integration pathway to enhance local capabilities. While developed in California, this modular framework is highly adaptable and particularly relevant for the Eastern Mediterranean. Applying dEPIC to existing telecommunications infrastructure could significantly enhance early warning capabilities for offshore hazards, such as those originating from the Cyprian Arc, where traditional station coverage is limited.