Università del Sannio, Benevento, Italy

Local microearthquakes were used to estimate site effects and source dynamic-scaling characteristics at Mt. Vesuvius, Italy. The selected data set is composed of low magnitude events (1.1 £ Md £ 3.6) recorded in 1996 and 1999 by nine digital shortperiod (1-Hz) seismic stations. Site response was evaluated by analysing data with three different approaches: 1) spectral ratios method of S-waves with respect to the average amplitude spectrum; 2) generalized inversion for site and source from the S-waves; and 3) generalized inversion from the coda waves. The results obtained with all three methods showed amplification of a factor of 1.5-2.5 in the 8-14 Hz frequency band for BKE and SGV sites and an amplification of 3 in a narrow band around 8 Hz for the CDT site. Method 2 allowed simultaneous determination of the source spectral shape for each earthquake. By assuming an w2 source model, we estimated the seismic moment Mo and corner frequency fc. The results show that most of the selected earthquakes are characterized by stress drops of 10 bars. The present results are encouraging for further investigation into the techniques for site-effect evaluation and for improving our knowledge of the scaling law of the source spectrum at Mt. Vesuvius.

Noise measurements were recorded using a dense short-period seismic array in Terzigno (Naples), a town that is located about 6 km from the Vesuvius crater. The aim of this study was to calculate a surface velocity model of the area under investigation through the application of the Spatial Autocorrelation (SPAC) method, with the hypotheses that ambient noise is stationary both in time and space, and that it is composed of surface dispersive waves. The correct knowledge of the surface structure is an important goal in site-effects studies. Correlation coefficients were calculated as functions of the azimuth on noise recorded at pairs of equally spaced stations in the frequency range of 1-8 Hz. Then, the spatial average correlation coefficients were compared to estimates over long-term recordings. The results appear to validate the hypothesis that ambient noise can be considered as a stochastic process. The correlation-frequency curves have been fitted to Bessel functions, from which the Rayleigh wave dispersion curve has been calculated. A velocity model has been derived from the dispersion curve using both trial and error and a standard inversion procedure. The results are consistent with those obtained from array measurements in the area in other studies (Scarpa et al., 2003).