Evaluation of microseismic array performances (EMAP) : case study of a deep metal mine monitoring network
Résumé
Underground mines can experience certain levels of microseismic activity as a consequence of stress perturbations and rock damages due to mining excavations. The understanding of the complex interaction between state of stress modifications and the generation of induced seismicity is a fundamental purpose in order to control the rate of seismicity and guarantee mine workers safety. In this context, microseismic monitoring in active mines has become a worldwide consolidated tool for observing and analyzing mining-induced seismicity. The performance of seismic networks, in terms of events detectability and location accuracy, is, thus, one of the basic concerns. However, a good performance of seismic networks is often hard to achieve, especially in the complex environment of active mines characterized by very low magnitude seismic events and where network geometries are limited by available galleries and mine production plans. In this work we present a simple methodology for the evaluation of microseismic array performances (EMAP). For a given network geometry in a monitored volume, EMAP determines: (i) the minimum events magnitude that can be detected and located and (ii) the distribution of predicted location errors for a specific magnitude. Given a velocity model and the attenuation relationship, for several virtual seismic sources regularly located within the monitored volume, EMAP estimates synthetic amplitudes, wave arrival times and polarization angles at each seismic probe. Based on the background noise at stations, signal-to-noise ratios (S/N) are computed per each couple source-receiver. The magnitude of completeness is, then, determined comparing the S/N ratios with imposed threshold for detection and location, respectively. In addition, for the estimation of expected location errors, initial synthetic values of arrival times and polarization directions are perturbed with errors, which follow a Gaussian distribution. Performing several simulations per each grid point, location errors are then given by the average value of the difference between the synthetic locations and the ones determined after input data alteration. We applied EMAP algorithm to the underground seismic network of the deep metal mine of Garpenberg (Sweden), which is equipped with both one-component and three-component seismometers, over a monitored area of 64×106 m3. Considering an empirical attenuation law for amplitudes, an homogeneous velocity model and determining the noise level at stations from real records, we tested the algorithm for different geometries of the monitoring network. Detectability performances were evaluated in the local magnitude range 0 ≤ ML ≤ -4.5, while location errors were determined for a ML of -2. The detectability performances are in agreement with magnitudes of real microseismic events normally recorded and located in the site. Results showed smaller location errors in the central area of the seismic network, which is around 1200 meters below ground surface, where source points are more constrained. Geometrical effects in detection and location performances are observed and caused by the heterogeneous locations of the stations due to exploitation constrains that prevent a complete optimization of the network. As expected, performances are improved, both in terms of events detectability and location accuracy, for network configurations when increasing the number of seismic stations. These tests showed that EMAP works properly not only for the performance evaluation of existing arrays, but also for defining efficient network geometries that may significantly reduce initial location errors in complex sites. Besides the application to local underground arrays our methodology can be also suitable for regional earthquake monitoring networks.