GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 3.1 43 method. Inversions are performed with the simultaneous iterative reconstruction technique (SIRT, Lytle et al. , 1978; Peterson et al. , 1985). Given the test disposition (reduced ray coverage due to difficulties in lowering the in-hole source) and the hole casing (possible interference of tube waves) the resulting seismic sections have to be considered with particular precautions. Nevertheless in both seismic sections (an example for P wave is reported in Fig. 1b) a shallow coverage of low velocity materials of about 2 m (top soil and highly altered/fractured granite, as also resulting from S1 and S2 core logs) is evidenced. A clear refraction boundary is noticed below this zone revealing an interface with a high-velocity but non-homogeneous formation. Intact granite velocity reaches about 2.9-3.1 km/s for P waves and 1.6-1.8 km/s for S waves. Estimated velocities appear to be quite low for an intact granite formation; direct measurement on rock samples are planned in this respect to obtain a direct verification. Two main low-velocity zones (about 2.2-2.4 km/s for P waves and 1.2-1.4 km/s for S waves) are also evidenced in both seismic sections within the massive granite formation. The position and dip of these zones agree with the fractures of the K4 system (50/75) whose traces are directly visible on the yard in front of the sanctuary. Low seismic velocities depths quite well compare with the outcomes of the stratigraphic log along the S1 sounding, enabling to associate the reduction in velocity values to zones of the rock mass with clear evidence of recrystallized fractures and fractures with alteration patinas. Given the accordance of all these evidences, the geophysical survey reveals a non-uniform (in shape, opening and velocity values) sliding plane probably associated with the presence of rock bridges whose ruptures could be the main cause of instability and which are the subject on which we focus the microseismic monitoring campaign. The microseismic monitoring network. The microseismic network has been installed at Madonna del Sasso in November 2013. It consists of four triaxial geophones (4.5 Hz) connected to a multichannel acquisition system (Granite - Kinemetrics, Inc.). Geophone location is reported in Figure 1a.The first two geophones are placed on sheltered areas of the high walls of the cliff, at a height of about 50 m below the top yard, on the south-eastern (ST1) and northern (ST2) side of the instable volume. The other two stations are placed in shallow manholes accessible from the panoramic square: ST3 is located in the lawn, near the inclinometric borehole S1, and ST4 is outside this unstable area, SE from the sanctuary. In the first operative weeks we managed to set up the best acquisition parameters for signal recording. We used a continuous recording at 250 Hz sampling frequency (later changed to 1 kHz in March 2014 for improving the first arrival time picking and obtain wider frequency content information) and a trigger recording based on a STA/LTA (Short Time Average over Long Time Average) detection algorithm. The STA/LTA ratio is continuously calculated and if it exceeds the user defined STA/LTA trigger threshold level, for any channel of the monitoring network, trigger is declared on the whole network. The network de-triggers if the STA/LTA ratio falls below another preset value (STA/LTA de-trigger threshold level). All calculations are made for every data sample in each of the twelve channels connected to the recorder, ensuring minimal time delay between the microseismic event and the triggering. Since these parameters of seismological interest are mainly used for the detection of earthquakes, we made several attempts to find threshold values suitable for the microseismic monitoring. We set a small STAwindow duration (0.3 s) since the shorter it is, the more sensitive to short and high frequency signals the STA/LTA trigger should be, and a LTAwindow duration of 30 s, with a STA/LTATrigger Threshold Level of 6. We also set a low value of STA/LTA de-trigger threshold (10%) to preserve complete coda waves for further analysis. Each channel was assigned a number of votes that it may cast towards getting the system to trigger. We selected the number of votes each channel would contribute (when it is triggered) to the total number of votes required to trigger the system (12). We gave zero votes to the six channels of ST3 and ST4 that we didn’t want to affect the triggering since they are located in