Long period soil amplification in the Po Plain (Italy) to account for site-effects in regional PSHA

Abstract

It is widely recognized that a significant proportion of the variability of earthquake ground motion is related to local geological conditions, which can modify the ground-motion amplitude, duration, and frequency. In particular, several investigations of deep sedimentary basins have highlighted that thick and soft sediments can strongly amplify long-period ground motion (> 1 s) (e.g., Anderson et al., 1986; Joyner, 2000; Milana et al., 2013; Massa and Augliera, 2013). The characteristics of long-period ground motion have gained growing interest in the civil engineering community because of the increase in the number of large-scale structures (i.e. tall buildings, long-span bridges etc.). In the framework of the site response estimation, both the S-wave velocity profile and the thickness of the soft sedimentary cover are considered fundamental parameters. At this regard, in a deep sedimentary basin at least three different kind of bedrock can be identified: geologic, engineering and seismic bedrock. The definition of bedrock is critical since it may provide very different reference conditions. The ‘geologic bedrock’ can be identified in correspondence with rock formations, whereas the ‘engineering bedrock’ can be identified based on the shear-wave velocity (Vs) value indicated in the current seismic code. Both the European and Italian seismic codes have defined the Vs transition from soft to stiff soil or rock (i.e., soil category A) at 800 m/s (European Committee for Standardization, 2004; Ministero delle Infrastrutture e dei Trasporti, 2018). In this sense, the Vs threshold of 800 m/s marks the top of engineering bedrock. However, this is a conventional value and it might not correspond to significant variation in the mechanical properties of the subsoil materials. On the other hand, the seismic bedrock is defined by a marked seismic impedance contrast between soft and hard soils (or rock), which results in a considerable ground motion amplification at the surface. In deep and wide sedimentary basins, geologic, engineering and seismic bedrocks do not always coincide, and the threshold of 800 m/s indicated in the current seismic codes (i.e., engineering bedrock) might not be significant for site amplification if does not also correspond to a seismic impedance contrast. In this sense, Mascandola et al. (2019) identifies the seismic bedrock of the Po Plain - one of the deepest and widest alluvial basin worldwide where several cities and critical facilities are present - in correspondence with a marked increase in the mechanical properties of the subsoil materials, which produces a measurable resonance effect at the surface in the medium-to-long-period range (i.e., 3.33 -1 s, that is 0.3-1 Hz). This corresponds to a marked seismic impedance contrast where the shear-wave velocity approaches, or exceeds, 800 m/s. In detail, to map the seismic bedrock depth we rely on an extensive collection of both existing and newly acquired ambient vibration measurements, with the aim of defining the soil resonance frequencies and the shear-wave velocity gradients within the soft sediments above seismic bedrock. Based on the collected data, an empirical regression model that relates the thickness of the soil deposits above the seismic bedrock to their resonant frequency is defined and applied to map the seismic bedrock depth in the Po Plain area. The resultant seismic bedrock map is correlated with the main unconformities recognized inside the Quaternary succession (Regione Emilia-Romagna,ENI–AGIP, 1998; Regione Lombardia, Eni Divisione Agip,2002). With the aim of providing a long period soil amplification model for the Po Plain, a regional shear-wave velocity model of soft sediments above seismic bedrock is performed through the interpolation of 54 S-wave velocity profiles selected after a quality check on the available data. The velocity gradients highlight two different zones inside the study area: one at Northwest and another at East-Southeast with higher and lower velocity gradients respectively. To compute the soil amplification functions, the velocity model is discretized into a 0.1° x 0.1° grid (~ 8 km x 11 km). For each grid node, a 1D soil model is defined, and a numerical ground response analysis is carried out to compute site amplification functions. The numerical model is verified at those sites with borehole seismic stations, where recordings of the same earthquake are available both at the surface and bedrock depth. Finally, with the aim of determining the influence of the deep geophysical discontinuities on the long-period hazard up to 3 s, the probabilistic seismic hazard assessment is computed at regional scale with the partially non-ergodic approach (e.g., Rodriguez-Marek et al., 2014; Kotha et al., 2017), and considering the most updated scientific improvements in seismogenetic zonation and ground motion prediction equations. The results will be compared with conventional PSHA estimations that accounts for site effects through the application of frequency-independent soil factors derived from seismic codes (European Committee for Standardization, 2004; Ministero delle Infrastrutture e dei Trasporti, 2018), or by adopting ground motion prediction equations (GMPEs) for specific ground types (Barani and Spallarossa, 2017).