Ente Nazionale per l'Energia Elettrica (ENEL), Roma, Italy

This article describes the results of a ground motion modeling study of the 1915 Avezzano earthquake. The goal was to test assuinptions regarding the rupture process of this earthquake by attempting to model the damage to historical monuments and populated habitats during the earthquake. The methodology used combines stochastic and deterministic modeling techniques to synthesize strong ground motion, starting from a simple characterization of the earthquake source on an extended fault plane. The stochastic component of the methodology is used to simulate high-frequency ground motion oscillations. The envelopes of these synthetic waveforms, however, are simulated in a deterministic way based on the isochron formulation for the calculation of radiated seismic energy. Synthetic acceleration time histories representative of ground motion experienced at the towns of Avezzano, Celano, Ortucchio, and Sora are then analyzed in terms of the damage to historical buildings at these sites. The article also discusses how the same methodology can be adapted to efficiently evaluate various strong motion parameters such as duration and amplitude of ground shaking, at several hundreds of surface sites and as a function of rupture process. The usefulness of such a technique is illustrated through the inodeling of intensity data from the Avezzano earthquake. One of the most interesting results is that it is possible to distinguish between different rupture scenarios for the 1915 earthquake based on the goodness of fit of theoretical intensities to observed values.

The strong motion accelerograms recorded during the 1990 Eastern Sicily earthquake have been analyzed to investigate source and attenuation parameters. Peak ground motions (peak acceleration, velocity and displacement) overestimate the values predicted by the empirical scaling law proposed for other Italian earthquakes, suggesting that local site response and propagation path effects play an important role in interpreting the observed time histories. The local magnitude, computed from the strong motion accelerograms by synthesizing the Wood-Anderson response, is ML = 5.9, that is sensibly larger than the local magnitude estimated at regional distances from broad-band seismograms (ML = 5.4). The standard omega-square source spectral model seems to be inadequate to describe the observed spectra over the entire frequency band from 0.2 to 20 Hz. The seismic moment estimated from the strong motion accelerogram recorded at the closest rock site (Sortino) is Mo = 0.8 x 1024 dyne.cm, that is roughly 4.5 times lower than the value estimated at regional distances (Mo = 3.7 x 1024 dyne.cm) from broad-band seismograms. The corner frequency estimated from the accelera- tion spectra i.5 J; = 1.3 Hz, that is close to the inverse of the dUl.ation of displacement pulses at the two closest recording sites. This value of corner tì.equency and the two values of seismic moment yield a Brune stress drop larger than 500 bars. However, a corner frequency value off; = 0.6 Hz and the seismic moment resulting from regional data allows the acceleration spectra to be reproduced on the entire available frequency band yielding to a Brune stress drop of 210 bars. The ambiguity on the corner frequency value associated to this earthquake is due to the limited frequency bandwidth available on the strong motion recordil1gs. Assuming the seismic moment estimated at regional distances from broad-band data, the moment magnitude for this earthquake is 5.7. The higher local magnitude (5.9) compared with the moment magnitude (5.7) is due to the weak regional attenuation. Beside this, site amplifications due to surface geology have produced the highest peak ground motions among those observed at the strong motion sites.

We used a set of previously published nonparametric attenuation functions to calculate average functions for the Italian region. These nonparametric functions describe the spectral amplitude decay of the S waves as a function of distance for 14 frequencies between 1 and 20 Hz for the regions of Lombardia-Piemonte, Eastern Sicily, Friuli, Marche and the Central Apennines. Since all the attenuation functions were obtained using the same methodology, we were able to make a fair comparison of the attenuation characteristics of the different regions. In general, while the Central Apennines show the strongest amplitude decay for all the frequencies analyzed, the regions of Lombardia and Eastern Sicily show the smallest attenuation. The Marche region also shows a strong amplitude decay, particularly for frequencies f > 3 Hz, and in the frequency band between 1 and 3 Hz the rate of decay of the spectral amplitudes with distance becomes similar to that of the region of Friuli. Since the attenuation functions analyzed represent different geologic and tectonic environments, we used them to calculate an average set of attenuation functions, one per frequency, for the hypocentral distance range between 10 and 120 km. The resulting functions permit a mean attenuation correction to spectral records of S waves in regions of Italy where the specific values of the attenuation parameters are unknown.