Integrated Deterministic and Probabilistic Strong Ground Motion Prediction: Application to 1980 Irpinia Earthquake, M=6.9, Southern Italy

The integration of probabilistic and deterministic approaches, generally adopted for computing scenario of past and future high magnitude earthquakes, is one of the strong motion seismology major tasks in recent years. Such an integrated technique allows overcoming some of the limitations inherently present in both the deterministic and probabilistic approaches. In particular: the estimate of b-value and seismicity rate are largely uncertain in the case of a single causative fault and/or fault system, empirical attenuation relationships (used to estimate the ground motion level at a given site) do not account for the details of the source/propagation process. On the other hand, deterministic scenario does not explicitly incorporate the frequency of occurrence of the earthquakes and its relation to the time of interest. Therefore, we propose a tool that integrates the probabilistic and deterministic techniques for scenario seismic hazard analysis. The deterministic approach is used for estimation of the ground motion parameters and their variability needed by the probabilistic technique. The b-value is assumed as the one computed from the Gutenberg & Richter law of the seismogenetic zone in which the causative fault is embedded whereas the seismicity rate is computed assuming the ``characteristic earthquake" recurrence model (Youngs and Coppersmith, 1985). The technique has been applied to the case of an earthquake having the complex source geometry of the 1980, M=6.9, Irpinia (Southern Italy) earthquake producing hazard maps at regional scale. As regards the deterministic modeling, main source characteristics of the earthquake are based on 1) previous studies (Bernard and Zollo, 1989; Cocco and Pacor, 1993), 2) modeling real strong motion data and observed intensities. The simulations are performed for a dense regional grid of receivers by hybrid k-squared source modeling technique. For this technique, the rupture process is decomposed into slipping on individual overlapping sub-sources of various sizes providing k-squared slip distribution. The hybrid simulation technique then self- consistently combines 1) the integral approach at low frequencies, based on the representation theorem and the k-squared slip distribution composed by the sub-sources, and 2) the composite approach at high frequencies, based on the summation of ground motion contributions from the sub-sources. Greens functions are computed by the DWN method in layered crustal model.