How to Assess the Moment Tensor Inversion Resolution for Mining Induced Seismicity: A Case Study for the Rudna Mine, Poland

Abstract

Underground exploitation of georesources can be highly correlated with induced seismic activity. In order to reduce the risk and improve the mining operations safety, the mining activity is monitored by a dedicated seismic network. Moment tensor inversion is a powerful method to investigate the rupture process of earthquakes in mines, providing information on the geometry of the earthquake source and the moment release. Different approaches have been proposed to estimate the source mechanisms, with some advantages and limitations. One of the simplest and most used methods rely on the fit of the polarity and amplitude of first P wave onsets. More advanced techniques fit the full waveforms and their spectra. Here, we test and compare moment tensor and focal mechanism estimations for both inversion techniques. In order to assess the inversion resolution, we built realistic synthetic data, accounting for real seismic noise conditions and network geometry for the Rudna copper mine, SW Poland. The Rudna mine pertains to the Legnica-Glógow Copper District, where thousands of mining induced earthquakes are detected yearly, representing a serious hazard for miners and mining infrastructures. We simulate a range of different processes and locations, considering pure double couple, deviatoric and full moment tensors with different magnitudes and located in different mining panels. Results show that the P-wave first onset inversion is very sensitive to the geometry of the seismic network, which is limited by the existing underground infrastructure. On the other hand, the quality of the moment tensor solutions for the full waveform inversion is mainly determined by the strength of mining tremor and the signal-to-noize ratio. We discuss the performance of both inversion techniques and provide recommendations toward a reliable moment tensor analysis in mines.