GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 3.1 47 Unfortunately, in other periods of winter frost the monitoring network was not correctly working. Conversely rainfalls (obtained from a near monitoring station ofARPAPiemonte) seem not to affect the rate of probable microseismic events. However, it is somehow appreciable in Fig. 3b that rainfalls affect the total number of recordings. Water flow and seepage, entrainment of debris in the fracture zones near the geophones can trigger the system to record a larger number of events. 3-D velocity model. After identifying microseismic signals among the large and various number of recordings we located them within the rock mass to constrain the most unstable zones subjected to fractures and understand the evolution related to source position identification along a fracture zone. At the current state, determining the hypocenter locations has been really challenging for several reasons: many channel recordings are very noisy and first arrival time picks are either lacking or inaccurate for some stations; even if we are monitoring a small volume, the seismic velocity inside the rock mass is highly heterogeneous, as it resulted from the geophysical characterization; there is no evidence of separated p and s onsets in the signals. To try to overcome the problem we have built a three-dimensional P-wave velocity model joining the DSM (Digital Surface Model) of the cliff obtained from a laser-scanner survey, the results of the cross-hole seismic tomography, the geological observations and the geomechanical measures of the most pervasive fracture planes (Fig. 1c). To do this the original DSM point cloud has been cleaned from vegetation and man-made structures, to obtain a rough DTM (Digital Terrain Model) of the granite cliff. The point cloud has been then resampled to a geometric cubic grid of points at 0.5 m spacing. The air velocity (338 m/s) was assigned to the points having an elevation higher than the DTM, and a constant velocity to the intact granitic rock mass (the higher velocity found from cross-hole tomography). With a GPS measurement survey, we then accurately georeferenced the geophysical test lines and the major fracture traces directly accessible from the yard. We extrapolated in depth the trend of the discontinuities either from the tomographic inversion (for the K4 system) or from the mean dip and dip direction of the systems. The equation of the planes that best fitted these points located on the discontinuities has been found and we gave to those surfaces the lower velocity obtained from cross-hole tomography. Now we have a quite accurate 3-D velocity model to be used for event localization (Fig. 1c). Tests with known seismic sources in accessible locations above the cliff are also planned in order to obtain a further direct verification of the used seismic velocities. Accordingly, to estimate the hypocenters of the microseismic events we will try to use the NonLinLoc (Non- Linear Location) software package of Lomax et al. (2000) for probabilistic, global-search earthquake location in 3D media. Conclusions. This paper shows the first results of an ongoing work concerning the geophysical characterization of an instable rock mass and the signals acquired from the microseismic monitoring network. The seismic survey allowed investigating in depth the fracturing state of the granite mass, in order to better understand the instability mechanism and to design the monitoring system. The outcomes allowed us to better investigate the presence and geometry of discontinuities, revealing a non-uniform sliding plane, with associated heterogeneous P and S-wave velocities, probably suggesting the presence of rock bridges whose ruptures could be the main cause of instability and can be monitored by the monitoring network. The obtained 3-D velocity model is the basic requirement for the processing and location of the microseismic signals. From the observation of the first months of monitoring data, first steps towards an adequate signal analysis and classification have been made obtaining a preliminary classification of recorded events. In the near future we will proceed to the localization of event sources, to the improvement and automation of data analysis procedures and to search for correlations between event rates and meteorological data, for a better global understanding of rock mass instability phenomena.