Preliminary estimation of attenuation properties in the High Agri Valley (Southern Apennines, Italy) by the coda attenuation method

High Agri Valley is an intermontane basin of the axial portion of Southern Apennines (Southern Italy), characterized by a very strong seismogenic potential. Indeed, a  Mw=7.0  earthquake occurred in 1857. Currently, the seismic networks managed by ENI Oil Company and INGV, installed in the area, continuously record a low-magnitude natural seismicity. Furthermore, two anthropogenic earthquake clusters are documented in two distinct sectors of the valley, located NE and SW of the artificial Pertusillo lake, respectively. The first cluster is related to the fluid-induced microseismic swarms caused by the injection, through the Costa Molina 2 well, of the wastewater produced by the exploitation of the Val d’Agri oilfield. The second cluster is due to a protracted reservoir induced seismicity (RIS) affected by the combined effects of the water table oscillations of the Pertusillo lake, the regional tectonics and likely the poroelastic/elastic stress due to aquifers in the carbonate rocks.

In this study we investigated the attenuation properties of the High Agri Valley crust by the estimation of the S-coda waves Qc^-1, as it is widely recognized the role of fluids on this parameter. We selected a dataset of about 1800 events acquired from 2001 to 2015 by the two above mentioned seismic networks, with local magnitude (ML) ranging from 0 to 3.3. We estimated the attenuation of the target area by means of a linear regression analysis of the amplitude decay curves of the envelopes of the seismograms; these were filtered in the frequency ranges centered on 1, 2, 3, 4, 5, 6, 8, 10, 12, 14, 16 Hz. The Qc estimates were performed by using different time windows for the envelope fitting, starting from the time T₁ to the time T_L (the lapse time). In detail, we adopted, as T_1, 1.0*Ts, 1.5*Ts and 2.0*Ts (being Ts the S wave arrival time), and as T_L 10s, 15s, 20s, 25s and 30 s from the event origin time. Only the components for which the condition T₁<T_L<T₂ was fulfilled were considered for the linear regression, being T₂ the end-time of the coda envelope; the latter was automatedly found by a proper methodology implemented in this study.

The obtained results show the increase of Qc as a function of f at all the considered T_L. Compared with other tectonic regions worldwide, in the High Agri valley the Q_c(f) is very low: the Q₀, that is the Q_c at 1 Hz, ranges between 8 and 57. At greater frequencies, the highest estimated Qc value is lower than 400. These evidences could be interpreted as the effect of fluids in the investigated crust, thus providing a further hint on their possible role in the seismicity of the area. A complete characterization of seismic attenuation of the High Agri Valley will require further investigations, that is the separation of scattering and intrinsic contributions in the total attenuation and a 3D imaging: indeed, the latter could highlight possible overlapping between spatial attenuation anomalies and seismicity distribution in the investigated area.