GNGTS 2022 - Atti del 40° Convegno Nazionale

212 GNGTS 2022 Sessione 2.1 ESTIMATING LOCAL EMPIRICAL SITE AMPLIFICATION MODEL FOR CITY OF LUCERNE IN SWITZERLAND P. Janusz 1 , V. Perron 1 , C. Knellwolf 2 , D. Fäh 1 1 Swiss Seismological Service, ETH Zürich, Zürich, Switzerland 2 Verkehr und Infrastruktur, Abteilung Naturgefahren, Kanton Luzern, Kriens, Switzerland Introduction. Site response evaluation is a crucial component of seismic hazard and risk estimation, especially in densely populated urban environments that are particularly exposed. Recordings of weak earthquakes can be used to obtain empirical ground amplification using among others the standard spectral ratio (SSR - Borcherdt, 1970) or Generalized Inversion Techniques (GIT - Andrews, 1986). However, in cities located in areas of low to moderate seismicity, it may take several years to record a sufficient number of high-quality events due to strong background noise. In addition, because of the high cost of deploying dense long- term seismic monitoring networks and the lack of free-space areas in the city, the empirical methods based on earthquake observations do not allow tomap site response with high spatial resolution. On the other hand, seismic ambient noise can be measured relatively cheap and fast even in the urban environment. However, as many authors show (e.g. Bonilla et al. , 1997; Field et al. , 1990; Perron et al. , 2018a), we are not able to estimate the correct amplitude of amplification function with methods based only on ambient noise such as noise-based spectral ratio (SSRn - Kagami et al. , 1982) or horizontal-to-vertical spectral ratio (HVSR - Nakamura, 1989; Nogoshi and Igarashi, 1971). Hence, we applied the hybrid SSR approach (SSRh - Perron et al. , 2018) that combines ambient vibration data and earthquake ground motion observation. Our study area is Lucerne in Central Switzerland which is a densely populated town located in a soft sedimentary basin. The seismicity in the area is low-to-moderate, however, there is historical evidence of several strong earthquakes with damage in the past (i.e. Mw 5.9 in 1601). In our project (Janusz et al. , 2022), we tested and tried to optimize the SSRh method and developed an amplification model in a broad frequency band for the area. This study is part of the URBASIS-EU project in the framework of the Horizon 2020 ITN. Data. A small temporary seismic monitoring network consisting of 10 velocimeters (3-component short-period seismometers Lennartz 5 seconds LE-3D 5-s with Centaur digitizers) was installed respectively in 2019 and 2020 for a few months to record weak local and teleseismic earthquakes. In total, the network was operating for about 1 year. In addition, three accelerometers of the Swiss Strong Motion Network (SSMNet - Hobiger et al. , 2021; Michel et al. , 2014) located in the city center were used as supplementary. In June 2020 and April 2021, we carried out two ambient vibration measurement campaigns with the same setup (LE-3D 5-s – Centaur), we recorded noise for at least 1-2 hours at 100 points. Moreover, about 200 hundred single-station recordings performed in the past were added to our dataset (e.g. Poggi et al. , 2012). Methodology. The SSRh method combines two methods earthquake-based (SSR) and noise-based spectral ratios (SSRn). In the first step (Fig. 1), a small seismic monitoring network consisting of a minimum of two stations where one is located on the rock and one in the basin has to be installed. Then, ambient vibration data has to be recorded simultaneously with earthquake monitoring stations allowing calculating SSRn between each site and basin- situated station. Finally, we correct the noise-based spectral ratio using the rock-relative SSR. To obtain SSR functions, we processed 44 local and teleseismic earthquakes recorded between November 2019 and April 2021. We considered signals from the P-wave until the coda wave arrival (Perron et al. , 2018b) with a signal-to-noise ratio (SNR) higher than three. Then, the geometrical mean was calculated from several realizations and smoothed using the