GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.1 213 Konno and Ohmachi (1998) algorithm with a b-value of 40. To derive the SSRn, we divided ambient vibration recordings into shorter windows and calculate the geometrical mean of the spectral ratios for all windows, and smoothed with Konno and Ohmachi (1998) algorithm with a b-value of 40. Because we installed more than one seismic monitoring station inside the basin, we obtained several SSRh functions for each point: where U is Fourier amplitude spectra of earthquake ground motion and u of ambient noise. For each site, we calculated the weighted geometrical mean of all realizations with weight ( w ) being the squared inverse of the difference of f 0 values between the site ( f 0 s ) and basin station ( f 0 b ): The final amplification functions are referenced to the local rock station, however, by multiplying them by the empirical spectral modelling (ESM) amplification function (Edwards et al. , 2013) for that local rock station, we can obtain amplification with respect to a Swiss reference rock profile (Poggi et al. , 2011). The ESM amplification functions are calculated routinely for all stations of the Swiss network. The final standard deviation is a joint geometric standard deviation of SSRn and SSR, and ESM amplification functions: Moreover, we mapped the fundamental resonance frequency f 0 for the area using all available ambient vibration recordings. We used the RayDec analysis (Hobiger et al., 2009) to retrieve the ellipticity of the Rayleigh wave and picked manually f 0 values. Summary of the results. In Fig. 2, we show the comparison between earthquake-based SSR functions and thehybridSSRhmethod for stationsof the temporary seismicmonitoringnetwork. Fig. 1 - Graph showing the principle of the SSR h method. Triangles indicate stations of the seismic monitoring network while circles are ambient vibration measurements. Capital U shows recorded earthquake ground motion and small u - ambient vibration data.