GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.2 303 SEISMO-STRATIGRAPHIC MODEL FOR THE MILAN (ITALY) METROPOLITAN AREA BY AMBIENT-VIBRATION MONITORING AND IMPLICATIONS FOR SEISMIC SITE EFFECTS ESTIMATION M. Massa, S. Lovati, R. Puglia, G. Brunelli, A. Lorenzetti, C. Mascandola, C. Felicetta, F. Pacor, L. Luzi National Institute of Geophysics and Volcanology (INGV), via Alfonso Corti 12, 20133, Milano, Italy A urban-scale seismo-stratigraphic model for the Milan (Italy) metropolitan area was developed based on a combination of seismic passive and active data supported by a dense grid of geological and stratigraphic information both from shallow and deep vertical wells (Massa et al. , 2022). The area, involving more than 3.0 milion people and a high density of industrial facilities, is located at the North Western limit of the Po Plain, a very large sedimentary basin located between the thrust belts of the Alps and the Apennines and characterized by a complex buried tectonic structures. The studied area is characterized by a low rate of annual seismicity (http://terremoti.ingv . it). In particular, in the last 40 years no local earthquakes with magnitude > 4.5 have occurred. However, the area has often undergone ground motion over the long-period, namely in the case of distant earthquakes with higher magnitude, such as the 2012 Emilia, Mw 5.8, seismic sequence (Luzi et al. , 2013). Long-period ground motion amplification is a significant issue in the Po Plain, where resonance phenomena at low frequencies have been well-documented by many studies (e.g. Luzi et al. , 2013; Milana et al. , 2013; Massa and Augliera, 2013; Paolucci et al. , 2015; Mascandola et al. , 2017). This work is an attempt to investigate the influence of the shallow to deep (down ~ 1.8 km) alluvial deposits for the Milan metropolitan area, where amplification effects in the frequency range 0.2 to 8 Hz are observed. Therefore, a detailed seismo-stratigraphic model obtained from the analysis of existing geological and stratigraphic data (e.g. Regione Lombardia, ENI- AGIP, 2002; GeoMol, 2015; Ispra 2016; ViDEPI project, 2009) and newly acquired geophysical data is proposed. In particular, new 37 single-station and 4 ambient-vibration arrays measurements were acquired in the area using seismometers with owner period of 5s, together with 4 active multichannel analysis of surface waves ( MAWS) . The passive microtremor arrays were performed with different geometrical configurations and a maximum aperture of about 600 m. To estimate the resonant frequencies of the soft sediments, the horizontal to vertical spectral ratio technique ( HVSR ) was applied to the ambient-vibration recordings, while to determine the Rayleigh-wave dispersion curves from passive array, the data were analyzed using the conventional frequency-wavenumber ( f-k ), the modified spatial autocorrelation ( mSPAC ) and the extended spatial autocorrelation ( ESAC ) methods. The array data were used to determine the local V S profiles via joint inversion of the Rayleigh-wave dispersion and ellipticity curves deduced from the HVSR . The results from HVSR show three main bands of amplified frequencies, the first in the range 0.15-0.25 Hz, the second from 0.4 Hz to 0.65 Hz and the third from 2 Hz to 8 Hz. A decreasing trend of the main peaks is observed from the northern to the southern part of the city, allowing to hypothesize a gradual increase of depth of the relative regional chrono- stratigraphic unconformities. The passive ambient noise array and MASW show a clear dispersion of the fundamental mode of the Rayleigh-wave in the range 0.4 Hz to 30 Hz, allowing to obtain detailed V s profiles with depth down to about 1.8 km. The results obtained fromambient noise HVSR are compared to the HVSR curves considering 17 earthquakes with magnitude from 3.7 to 5.8 recorded by the IV.MILN station (http://terremoti.ingv.it/instruments/station/MILN ) in the last 10 years. The Vs profiles are finally validated through 1D numerical models performed assuming a linear elastic behavior of the soil using the SHAKE91 code (Idriss and Sun, 1993).