Using ambient noise <i>k</i>0 estimation to improve microearthquake source parameters assessment in the High Agri Valley

SUMMARY Characterizing the sources of small earthquakes allows us to explore the physical mechanisms involved in earthquake nucleation and provides insights into the deformation and stress propagation processes that trigger ruptures; understanding the properties of the sources is also crucial for assessing the seismic risk associated with seismicity induced by human activity. Within this context accurate estimation of source parameters for low-magnitude events is challenging, mainly due to the significant impact of high-frequency attenuation, which acts like a low-pass filter on seismic spectra of microearthquakes, leading to a possible underestimation of the true corner frequency. To limit these effects and to improve the accuracy of the source parameters for microseismicity, the k0 parameter is included in the spectra inversion. We focus on the microseismicity that occurred in the High Agri Valley (southern Apennines, Italy), an area affected by both natural and induced microseismicity. The high-frequency energy decay is modelled by estimating, at each station, the parameter k0 from the ambient seismic noise to avoid the trade-off between the high-frequency attenuation and the stress drop of the earthquake source. We estimated the k0 parameter using noise displacement spectra in the 15–40 Hz frequency range at eight broad-band stations belonging to the High Agri Valley Geophysical Observatory. Our results indicate that k0 varies between 0.02 and 0.04 s at the considered stations, with well constrained values in the selected frequency range. The estimated k0 values are integrated into the SourceSpec code to correct the high-frequency displacement amplitude spectra for the near-site attenuation when calculating the source parameters. Our procedure is tested on 72 microearthquakes (0.4 ≤ MW ≤ 2.7) of the Castelsaraceno sequence that occurred in August 2020 in the southwestern sector of the High Agri Valley. The results show the cut-off frequency of 10 Hz observed in uncorrected spectra of earthquakes with MW &amp;lt; 2.0, is no longer recognizable, and the threshold to estimate reliable corner frequency decreases by almost an order of magnitude, from MW = 2.0 to MW = 1.2. Moreover, events with MW &amp;lt; 1.5 also align with the 0.1 MPa stress-drop curve after k0 correction, and the fc–M0 scaling approaches conditions close to self-similarity. We therefore propose that smaller earthquakes with MW ≈ 1.5 may follow a different physical mechanism, as they exhibit stress drop values around 0.1 MPa - an order of magnitude lower than those of events with MW &amp;gt; 2.0.