GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.2 373 on the retaining cable, together with a bit of luck, the sensor is leveled at its best. Anyway, the tilt tolerance of the downhole sensor is ±2.5°. In some cases, the conditions of a pre-existing wells may suggest to locate the posthole sensor along the casing, rather than the bottom. To face these situation, an electro-mechanical positioning system is under development. It is based on two curved metal plates mounted on an electrical powered jack, which also serves as base for the sensor (Fig. 1e). The system has been designed to be with the center of mass aligned with the upper anchoring, to ensure a levelling within the tolerance limits. An identical post-hole sensor is then located at surface: it will serve as a reference for the orientation of the down-well sensor. This sensor will be later replaced (weeks to month) with an accelerometer, then, the site in its final configuration is integrated within the rest of the earthquake monitoring network The horizontal axes of sensors are normally oriented along the north and the east directions using a magnetic compass, but this operation is not possible for posthole installation. It is of fundamental importance to determine the orientation before to integrate its data within the network. There are many studies establishing the orientation of a borehole instrument using different methods, such as receiver functions, body-wave polarization studies, surface-wave polarization studies, ambient seismic noise, and seismic noise from ocean waves (see Lin et al. , 2022 for an overview). In the present study, we use an approach based on the cross-power spectra of the seismic noise between the two sensors. The cross-power spectra is determined between horizontal components of the surface reference sensor and the posthole sensor. The horizontal components of the posthole sensor are rotated by applying a rotationmatrix. The rotation was exacted in steps of 1°; the cross power spectrum is then calculated at angular step in the whole range of available frequencies (120s - 50 Hz). For the purposes of estimating the correct horizontal orientation of the sensor in the well, only low frequencies (<1 Hz) are considered. Fig. 2 shows the average value of the cross power spectrum in the 120s - 1 Hz range, as a function of the rotation angle. The maximum value of the average cross power spectrum identifies the best estimate of the rotation angle. The closer the maximum value is to unity, the more robust is the estimate of the rotation angle. The estimations of the angular correction are quite reliable (Fig. 2) and in the case of FAVA site it has been further refined using the signal of a teleseism. Fig. 2 - Estimation of the angular deviation of the pairs of horizontal components of the two sensors (i.e. surface and down-well). The vertical dashed lines correspond to the maximum values of the coherence and indicate the angular correction (21° for LIBRI and 349° for FAVA). The analyses of the early signals show some anomalous waveforms which have been interpreted as the results of micro cracks consequent to the stabilization of the sandy bed which surrounds the sensor. The occurrence of such oscillations decreases with time and they disappear after some weeks. The reduction of the background noise has been evaluated comparing the power spectral densities of the surface and posthole sensors (Fig. 3). Below 3 s, and down to 120 s, the reduction of the power is more than 10 dB. This reduction at very