No abstract

[1] During three moderate-magnitude earthquakes occurred in September-October 1997 in the central Apennines, Italy, accelerations larger than 0.5 g were recorded in the town of Nocera Umbra, 10 to 15 km N-NW of the epicenters. The accelerograph is sited in a fault zone, close to a N30°E tectonic contact. Six temporary seismological stations installed across the fault recorded 82 aftershocks occurred in two seismogenic zones: the Colfiorito-Sellano area, S-SE of the array, and the Gualdo Tadino area, to the north. The array data reveal large variations in terms of both peak ground motions and spectral amplitudes. Within the fault zone, amplifications show a strong dependence on the source azimuth. At the accelerograph site, the effects are particularly large for events from S-SE: peak ground motions are a factor of 14 larger than those of a reference site and conventional spectral ratios attain amplitudes as large as 50 at 7 Hz along the N30°E direction of motion, parallel to the strike of the fault. Nineteen strong motion accelerograms were then used to compare ground motion properties between weak and strong events up to M 0 = 1.2 Â 10 25 dyn cm. A particle motion analysis shows that the directional effect is also present in the strongest motions, even though the amplification of peak ground motion decreases when M 0 increases. Results from stochastic simulations indicate that such a behavior is not due to nonlinearity: applying the empirical weak motion transfer functions in a purely linear model the observed peak ground motions of the largest events are fit satisfactorily.

[1] The paper presents the results of a case study conducted on the Holocene alluvial deposits of the Tiber River valley, in the city of Rome. The main test site selected for the study, Valco S. Paolo, is located about 2 km South of Rome's historical centre. The alluvial deposits were dynamically characterized in a comprehensive way via site investigations and geotechnical laboratory tests. Normalized shear modulus decay and damping curves (G/G 0 and D/D 0 vs g) were obtained for the dominantly fine-grained levels. The curves demonstrate that these levels have a more marked shear stiffness decay if compared with the underlying Pliocene bedrock. Decay curves from laboratory tests for the Tiber alluvia correlated well with the trend of the function proposed by Hardin and Drnevich, making it possible to derive their specific interpolation function coefficients. Use was made of the extrapolation of the findings from the Valco S. Paolo test site to a large part of Rome's historical centre by means of two other test sites, supported by an engineering-geology model of the complex spatial distribution of the Tiber alluvia. The experimental Valco S. Paolo Vs profile was extrapolated to the other test sites on the basis of a stratigraphic criterion; the analysis of seismic noise measurements, obtained for the three test sites, validated the engineering-geology based extrapolation and showed that the main rigidity contrast occurs inside the alluvial body (at the contact with the underlying basal gravel-level G) and not between the alluvia and the PlioPleistocene bedrock, composed of highly consistent clay (Marne Vaticane). The 1D modeling of local seismic response to the maximum expected earthquakes in the city of Rome confirms that the deposits have one principal mode of vibration at about 1 Hz. However, the simulation also evidenced that the silty-clay deposits (level C), making up the most part of the Tiber alluvial body, play a key role in characterizing the soil column deformation profile since it can be affected by non linear effects induced by the maximum expected earthquake when some stratigraphic conditions are satisfied.

S U M M A R YDuring the M W 5.7 and 6.0 Umbria-Marche earthquakes of 1997 September 26, the historical centre of Nocera Umbra suffered MCS intensity VII-VIII. The zone is located on the top of a hill, a condition potentially favourable to ground motion amplification. However, also vulnerability is higher on the hill because of the ancient age of buildings. A temporary array of eight seismological stations was installed across the hill to quantify the amplification effect due to topography. Waveforms of 14 aftershocks (2.6 < M L < 4.1) are selected for the analysis. During each earthquake the largest amplitudes are observed on the hilltop, spectral ratios are computed using rotated horizontal components to search for directional effects. Amplifications are found in two separate frequency bands: one in the range 2-4 Hz, where the increase of amplitude is moderate (never exceeding a factor of 4) and the polarization is transversal to the hill major axis; the second above 10 Hz, where amplifications are larger and reach values as high as 25 Hz. High-frequency polarization varies for different sites and frequencies suggesting that smaller-scale complexities control the high frequency response. Synthetic seismograms of 2-D models confirm the occurrence of amplification, although not all details are fit by numerical simulations and the agreement between observations and models is significant only in terms of the fundamental resonance frequency, around 3 Hz. In the models, amplifications are much smaller than the observed ones. We conclude that topography could have been responsible for a small increase of damage in the hill zone but the most significant role on damage was played by the locally higher vulnerability.

[1] In winter 1998 a small-aperture ($200 m), four-station array was operating in the middle of the Colfiorito plain. Waveforms of seven small magnitude (2.1 < M < 3.7) local earthquakes are analyzed in this study. The peculiarity of the array seismograms is a $2-min long duration of the horizontal ground motion. The predominant frequency in the plain is around 1 Hz. In this frequency band, earthquake-induced ground shaking is synchronous for a few seconds during direct S waves; after this time window, wave trains show an increasingly chaotic behavior within the array. The energy variation of the horizontal ground shaking between the array and a nearby rock outcrop exceeds a factor of 500 at 1 Hz. In order to help understanding of observations and construct models for future numerical simulations, the buried structure of the Colfiorito plain has been investigated through seismic refraction profiles and geoelectric measurements. A three-dimensional reconstruction of the basin structure reveals an extremely complex bedrock topography, which is probably a recurrent feature for intermontane basins in tectonically active regions. Apparent velocities and backazimuths in the frequency band 0.5-1.5 Hz indicate that a 180-m deep depression NW of the array is responsible for the generation of edge-diffracted surface (Love) waves. Their arrival to the array breaks the synchronism of vertically reverberating waves which predominate in the first seconds of the S-wave window.