The site-corrected source scaling pattern is estimated for local earthquakes (0.9≤MD≤3.6) at Mt. Vesuvius. The dataset comprises 35 low-to-moderate local earthquakes recorded by 14 three-component seismic stations during 1993, 1996, and 1999.
Site-transfer functions in the frequency range 1 Hz–25 Hz are estimated from the spectra of S waves and coda waves and from the horizontal-to-vertical (H/V) spectral ratios. We applied the direct spectral ratios method to S waves, considering as a reference the average spectrum and the inversion method to S waves and coda waves. The site amplification on the coda waves was also compared with that evaluated using the wavelet transform. The standard deviation associated with the experimental results is computed for all of the used methods.
Results indicate a general agreement among the methods, and the site-transfer functions show interesting features. The highest amplifications are found for frequencies lower than 12 Hz for sites located at lower altitude. The methods based on coda waves show highest amplification with respect to the methods based on S waves for most of the sites located in the summit part of the volcano. This can be a phenomenon of coda localization, which consists in the trapping inside the upper part of the volcano of scattered waves. The H/V spectral ratios do not show total agreement with the other methods, mostly for the sites located in the summit part of the volcano. The discrepancies among the results obtained in this work are also due to the different normalization applied in the methods of analysis.
Generalized inversion method allowed us to estimate the source scaling of the site-corrected source seismic spectrum for the investigated area. The source scaling obtained in terms of seismic moment and source radii shows that the seismicity of Mt. Vesuvius is characterized by stress drop as low as a few bars (10 bars) except for the event of (). The scaling pattern shows an apparent linear relationship between source size and seismic moment (for MD≤3.3) but the statistical test shows that the linear trend has low reliability.