Temporal and spatial evolution patterns of earthquake clusters in the southern Sichuan industrial mining zone—Advanced analysis based on multisource data fusion and machine learning

The southern Sichuan industrial mining area is an important region of shale-gas hydraulic fracturing. It has experienced an exponential growth in seismic activity since 2016. It is one of the most active earthquake swarm zones on the eastern margin of the Sichuan-Yunnan block. This study addresses the scientific challenge of intelligently extracting spatiotemporal evolution laws from massive seismic data by employing multisource data fusion and artificial intelligence-enhanced analysis to reveal the spatiotemporal evolution and migration patterns of seismic activity under mining disturbances. The three main aspects of this work are as follows: (1) a coupled tectonic-engineering analysis was performed to analyze the tectonic controlling factors of the earthquake distribution by integrating surface fold morphology, basement burial depth zoning, spatial configuration of the Silurian Longmaxi Formation mining layers, and high-resolution seismic reflection profiles. (2) Through intelligent identification of earthquake clusters using machine learning algorithms such as HDBSCAN clustering and K-nearest neighbors supervised classification, 31 typical earthquake clusters were extracted from precisely located seismic catalogs, and spatial vulnerability zones and their dynamic migration patterns were quantitatively characterized. (3) The spatiotemporal characteristics were decoupled by combining time-series clustering to isolate interference events, and an evolution model of “spatial morphology-temporal clustering” of earthquake clusters was constructed. This model revealed the synergistic driving mechanism of mining stress and regional tectonic stress. The results showed that earthquakes are concentrated in fold structural units with weak control by the known faults in the Changning Anticline and the basement junction zone on the southern flank of Weiyuan. The earthquake clusters exhibit directional migration southeastward in the Weiyuan-Changning block and southwestward in the Rongchang-Luxian block, ultimately forming a new earthquake-prone area in Luxian-Da’an. The spatiotemporal evolution model of earthquake clusters offers a decision-making framework for long-term regional economic development planning and oil and gas extraction management.