PSI - Issue 78

Iunio Iervolino et al. / Procedia Structural Integrity 78 (2026) 1553–1560

1555

of the chosen GMM. Then, the occurrence of major earthquakes on the source is modelled via stochastic models reflecting the seismic history on the source, calibrated based on the fault’s known history, from geology and geophysics. The remainder of the paper is structured so that the basics of near-source probabilistic seismic hazard analysis are recalled, together with the procedure and models chosen to describe the rupture and consequent ground motion at the site. The results in terms of mean spectral ordinates, conditional to the occurrence of one event on the source, are presented and discussed. Subsequently, the considered models for earthquake occurrence with memory of the source history are introduced and their calibration described. The probability of the occurrence of the events from these models are compared to what would be predicted by an HPP. Some final remarks conclude the paper. 2 NS-PSHA conditional on earthquake occurrence The near-source PSHA discussed herein aims to account for ground motion at the site of Messina, possibly exhibiting forward-directivity velocity pulses. This NS-PSHA is carried out assuming (conditional on) the occurrence of an earthquake of unspecified (i.e., unknown) magnitude and hypocentral location on the ITIS013 fault (Fig. 1). The analysis considers (5% damped) spectral pseudo-acceleration, ( ) , where is the natural vibration period, as the ground motion intensity measure. The parameters of this east-dipping normal fault are obtained from DISS, which proposes strike, dip, and rake angles equal to 30°, 29°, and 270°, respectively. The (moment) magnitude associated by DISS to the ITIS013 fault is =7 , with corresponding rectangular geometry having a 40×17km 2 surface projection and minimum/maximum depths of 3 km and 12.7 km, respectively. Nevertheless, in the analyses, magnitude is modelled as a random variable (RV), uniformly distributed in the 7±0.3 interval; see Fig. 2a. Because the rupture area scales with magnitude and to accommodate this magnitude uncertainty, this preliminary analysis assumes that ruptures can occur over a wider area with respect to ITIS013, taken as a 60×18km 2 rectangle (surface-projected) that approximately coincides with the composite source ITCS016 from DISS.

Fig. 2. Probability density function of moment magnitude (a); finite-fault geometric parameters seen from above (b) and in cross-section (c). The Messina Strait fault considered in this study is a normal fault. To perform NS-PSHA for dip-slip ruptures, the parameters needed to define finite-fault geometry are treated as RVs. The rupture geometry RVs considered are: the rupture area ( ), the along-strike and along- dip distances of the rupture plane from the fault’s western and bottom edges, and , and the along-strike and along-dip distances of the hypocenter from the rupture edges, ℎ and , all shown in Fig. 2b and Fig. 2c. In the notation that follows, the last four RVs are grouped in the random vector = { , , , ℎ } . Assuming that all ruptures on the fault will be rectangular and maintain a ratio of along-strike ( ) to along-dip ( ) dimensions ⁄ = 2 , a realization of rupture area = and of the vector = { , ℎ, , ℎ } , univocally defines the rupture plane and its location. Under these assumptions, the probability that an ( ) threshold, , is exceeded at the site, that is, [ ( ) > ] , can be computed, according to Chioccarelli and Iervolino (2013), as in Equation (1), where [ ( ) > ∩ ] is the joint probability that, given earthquake occurrence, the event produces ground motion at the site that exceeds without being pulse-like, while [ ( ) > ∩ ] is the probability that ( ) > with the shaking exhibiting impulsive characteristics: