Surface ground-motion characteristics and path-corrected PGV prediction for mining-induced seismicity: a case study

Strong mining-induced earthquakes in deep mines threaten surface infrastructure, yet their surface vibration fields are difficult to predict because propagation paths cross goafs, fractured overburden, and variable Quaternary cover. This study used a dense surface monitoring array at the Dongtan Coal Mine to characterize the time–frequency evolution and spatial attenuation of typical mining-induced seismic events. Goaf-path length and Quaternary cover thickness were incorporated into a conventional distance-attenuation relationship to build an improved peak ground velocity (PGV) prediction model. The results show that surface waveforms change from short, high-amplitude near-field pulses to multi-peaked coda wave trains in the mid-to-far field. This evolution is marked by envelope broadening, delayed peak arrival, spectral migration to lower frequencies, narrowed bandwidth, and increased spectral decay. Propagation through fractured goaf zones and thick shallow cover accelerates high-frequency attenuation and produces strong azimuthal differences in PGV. Compared with the distance-only model, the path-corrected model improved the coefficient of determination from 0.79 to 0.87 and reduced RMSE and MAPE from 0.38 cm/s and 24.0% to 0.09 cm/s and 21%, respectively. The model is empirical, but it illustrates how path and site effects shape mining-induced surface vibration. The representative M L 1.9 event remained below safety-control thresholds for major surface structures. The proposed framework supports rapid post-event assessment and refined risk zoning in deep mining areas.