The use of PhaseNet for Event Identification of Microearthquake Monitoring in Geothermal Field

Geothermal energy is a sustainable energy source that requires continuous microseismic monitoring to assess reservoir integrity and geomechanical behavior. Traditional phase identification methods are challenged by noisy environments and complex waveforms, especially in geothermal fields. This study explores the efficacy of PhaseNet, a deep learning neural network model, in detecting P and S wave arrival times for micro-earthquake events. The PhaseNet model was retrained using local seismic data from a geothermal field and tested for its performance in identifying seismic phases. The results were validated against a manual seismic catalog, with additional clustering and association analysis conducted using GaMMA and hypocenter locations determined with NonLinLoc. The findings demonstrate that PhaseNet, combined with GaMMA, provides robust phase detection capabilities, essential for early-stage monitoring in geothermal development.