SYNTHETIC SEISMOGRAMS FOR SEISMIC HAZARD ASSESSMENT: CASE-STUDIES IN ITALY

The ground shaking prediction for future earthquakes is the key issue of any assessment of seismic hazard. The most straightforward way to produce ground shaking scenarios makes use of Ground Motion Prediction Equations (GMPEs). The employment of the GMPEs, however, becomes questionable in near-fault conditions. Indeed, empirical models, even though calibrated on past records of acceleration waveforms, generally provide a simplified representation of the ground motion predicted around the source. This is essentially due to the paucity of data for moderate-to-large events at short source-to-site distances. When a more detailed and realistic representation of the seismic shaking is required, numerical simulations can be a reliable alternative. Several simulation techniques have been proposed to this aim with different levels of complexity depending on the approximations introduced to represent seismic source, wave propagation and site response. The result is that the modeling of the same event with different techniques may produce ground motions differing in terms of amplitude, duration and frequency content. For this reason, before using numerical simulations to predict ground motion of future earthquakes, the adopted methods should be carefully validated through quantitative procedures based on the comparison between observed and simulated waveforms. One of the goal of the S3 project (INGV-DPC Agreement 2005- 2007), coordinated by Francesca Pacor and Marco Mucciarelli, was the generation of shaking scenarios at bedrock and including site responses. Starting from these pioneer studies, several issues related to the use of synthetics seismograms for hazard assessment have been addressed. In this work, we present an overview of the modeling studies of the major Italian earthquakes (1980, M 6.9 Irpinia event; 1990, M 5.8 Sicily earthquake; 1997, M 6.0 Umbria Marche main-shock; 2009, M 6.3 L’Aquila events), carried out applying different simulation techniques (stochastic, deterministic- stochastic, broad-band). We also discuss some examples of scenario-events, generated at single and multiple sites, illustrating our approaches to handle the synthetic ground-motion variability. Finally, we show an hazard evaluation based on hybrid ground motion prediction equations, calibrated using synthetics waveforms at short distances and recorded waveforms, far from the source