Graph Neural Networks for Microseismic Event Detection: Focusing on Distributed Acoustic Sensing Data

Microseismic event detection is paramount for surveilling subsurface activities including hydraulic fracturing, geothermal energy extraction, and evaluating seismic hazards. Distributed Acoustic Sensing (DAS) technology provides high-resolution spatial and temporal datasets, rendering it an indispensable instrument for identifying microseismic events. Nonetheless, the substantial volume of data and the intricate spatiotemporal interrelations inherent in DAS data present considerable obstacles to conventional signal processing techniques. This research presents a novel approach utilizing Graph Neural Networks (GNNs) for microseismic event detection in DAS data. By displaying the spatial and temporal relationships between DAS sensing points as graph structures, the GNN identifies the underlying spatiotemporal patterns linked to microseismic events. According to empirical results, the suggested approach outperforms conventional machine learning frameworks, achieving higher detection accuracy and more computational efficiency.