Journal of Seismology 4: 555–565, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 555 Study of site effects in the area of Nocera Umbra (Central Italy) during the 1997 Umbria-Marche seismic sequence A. Caserta, F. Bellucci, G. Cultrera, S. Donati, F. Marra, G. Mele, B. Palombo & A. Rovelli Istituto Nazionale di Geofisica, Rome, Italy Received 15 September 1999; accepted in revised form 17 March 2000 Key words: Nocera Umbra , site effects, weak motions Abstract During the M L = 5.6 and 5.8 earthquakes occurred in central Italy on 26 September 1997 the historic centre of Nocera Umbra, lying on top of a 120 m high hill, was diffusely damaged (VII-VIII degrees of MCS intensity). Some recently built houses in the modern part of the town suffered an even higher level of damage. A temporary seismic array was deployed to investigate a possible correlation between local amplifications of ground motion in this area and the observed pattern of damage. After a geologic and macroseismic survey, eight sites were selec- ted as representative of different local conditions, such as topographic irregularities, sharp hard-to-soft lithology transitions, alluvium-filled valleys, and both undisturbed and deformed rocks. Horizontal-to-vertical spectral ratios for both microtremor and earthquake recordings, as well as spectral ratios referred to undisturbed rock sites, were used to quantify local variations of ground motion. In spite of the diffuse damage in the historic centre of Nocera Umbra, a small amplification is observed at the stations on the hill’s top. This suggests that the higher vulnerability of the ancient buildings mainly accounts for the diffuse damage in that part of the town. In the frequency band of engineering interest (1 to 10 Hz) the largest amplifications of ground motion are found at soft sites: in the Topino river valley, where many episodes of severe structural damage occurred, spectral amplification is significant over a broad frequency band ranging from 2 Hz to more than 20 Hz. In particular, in the central part of the valley high amplification (> 4) is found from 3 to 10 Hz, reaching a maximum of 20 around 4 Hz. At the edge of the valley, close to the soil-to-rock transition, amplification is as large as 10 in a frequency band ranging from 4 to more than 20 Hz. A significant amplification (by a factor of 10 around 10 Hz) is observed also at one of the rock sites, possibly due to the presence of a cataclastic zone related to the activity of a regional fault that altered the mechanical properties of the rock. Introduction On 26 September 1997, two moderate earthquakes (M L = 5.6 at 00:33 GMT, and M L = 5.8 at 09:40 GMT) occurred close to the Umbria-Marche border (Fig- ure 1). Seismic activity continued until middle 1998 with several hundred events per month. The inset of Figure 1 shows the epicentres of the two main shocks with their focal mechanism, aftershock distribution, and the most severely damaged towns. In spite of the small distance between the two mainshocks (less than 5 kilometres) and the similar focal mechanism, these earthquakes produced a different distribution of the largest damage. While the first shock caused an MCS intensity of VII-VIII degrees at the closest villages of Annifo, Cesi, Colle Curti, the second shock produced the largest damage northward. During the 09:40 GMT event the area of Nocera Umbra felt a maximum MCS intensity of VIII-IX degrees (Tosi et al., 1999). The largest effects felt in Nocera Umbra during the M L = 5.8 event can be ascribed to a source effect due to the northward directivity inferred by Pino et al. (1998) from broad-band waveform modelling. However, a local amplification of ground motion at Nocera Umbra was already observed by Rovelli et al. (1988) using strong-motion accelerograms of previous earthquakes. These authors observed a significant amplification in a frequency band ranging from 4 to 10 Hz through the