Semi-automatic detection and localization of microseismicity induced by a "salt dissolution provoked" cavity collapse
Résumé
Natural underground cavities, active or abandoned mine workings, particularly when they are shallow, can provoke large scale land subsidence and collapses attended by catastrophic social-economic impacts. The potential of passive microseismic monitoring to prevent such disasters was already indicated by several studies. Nonetheless, to further improve monitoring reliability a better understanding of associated seismicity is inevitable. In this context, within a large multi-parameter research project at Cerville-Buissoncourt in Lorraine, France, the growth of a single, shallow, about 150 m diameter salt cavity created by salt dissolution mining was surveyed from 2004 until 2009 when the cavity reached its critical size and a 'controlled' collapse was initiated. During the experiment, a large microseismic data set was recorded by a triggered, high resolution geophone monitoring system. Initial processing and data inspection reveal very unusual seismic signals mainly appearing in complex swarming sequences (Mercerat et al., 2010; Contrucci et al., 2011). To resolve spatio-temporal characteristics of associated seismicity we developed an semi-automatic seismic event detection and localization algorithm adressing these abnormal signal characteritics. The detector design is based on a spectral envelope function calculated for each seismogram. By this function, coherent signals are distinguished from signals comprising rather randomly distributed frequency proportions as noise or CODA waves. First application tests demonstrated highly improved event detection results when analysing seismic events of highly varying size and duration occurring in a swarming sequence. In addition, we localized the detected seismic events using inter-station amplitude ratios as introduced by Battaglia and Aki (2003) and Taisne et al. (2011). Within this approach, hypocenter source inversion relies on the decay of seismic wave amplitudes along the source-receiver path. As a result, no troublesome a priori phase segmentation is needed and the entire data set can be processed. To calibrate the local seismic attenuation law we used ~700 seismic events with known hypocenter locations found by previous studies (Klein et al., 2011). The final optimized localization algorithm sufficienty constrained the tendency of actual hypocenter source location in the cavity region. Taken all together, our detection and localization strategy provides an appropriate first order approximation to study spatio-temporal attributes of seismicity from huge data sets associated with seismic signals of unknown or complex signature as observed for Cerville-Buissoncourt.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...