Spatiotemporal clustering of microseismic signals in mining areas: A case study of the Baoji lead‑zinc mine in Shaanxi, China

Abstract

Microseismic activity is a critical indicator of stress redistribution, geological anomalies, and potential hazards in underground mining environments. Traditional clustering methods, however, often fail to capture the complexity of spatiotemporal distributions and the diverse triggering mechanisms of mining-induced microseismic events. To address this gap, we propose a novel clustering framework that combines K-means and Gaussian Mixture Models (GMM) to improve the classification and understanding of microseismic signals. Using a dataset of over 5000 high-quality events from the Shaanxi Baoji Dongtangzi lead–zinc mine, we establish a dynamic completeness magnitude threshold (m ≥ −1.0), ensuring the reliability of the seismic dataset. Our analysis reveals distinct spatiotemporal patterns, magnitude distributions, and spatial clusters, driven primarily by geostress redistribution, mining operations (e.g., blasting, drilling, ore transportation), and noise. The time-interval analysis further demonstrates non-Poisson clustering behavior, reflecting the impact of stress redistribution and operational schedules on microseismic activity. The results not only deepen the theoretical understanding of mining-induced seismicity but also offer practical insights for optimizing risk management and enhancing safety protocols in underground operations. Additionally, this approach provides a scalable framework for broader applications in geologically similar mining regions, contributing to safer and more efficient resource extraction practices worldwide.