Detecting Microseismic Events Using Cross-Fault Borehole DAS

Microseismic events account for the majority of seismicity, however, sparse station spacing hinders the detection of such small events. In recent decades, distributed acoustic sensing (DAS) has shown its power to provide a denser spatial sampling in an array sense, to resolve weak signals that are often missed by conventional seismometers. In eastern Taiwan, the Chihshang fault plays a key role in accommodating deformation along the Longitudinal Valley fault system, where frequent small earthquakes and fault creep occur. In this study, we develop a new workflow for microseismic event detection by integrating borehole DAS data with the deep-learning-based automatic phase picking model PhaseNet. An event is declared when more than 75% of channels record P-wave picks and more than 30% record S-wave picks within a 1-s time window. We analyzed three months of DAS data from March to July 2025. As a result, we identified approximately twice as many events as those reported in a deep-learning-based earthquake catalog constructed using only surface seismic stations. These results suggest that borehole DAS provides an effective complementary constraint for detecting earthquake-generated wave trains. This processing workflow can significantly improve the detection capability for microseismic events, leading to higher seismic catalog completeness and finer fault structure near the Chihshang region.