The 6.0 Mw earthquake that hit central Italy on 24 August 2016 caused at Amatrice damage of unprecedented severity for an earthquake of such a magnitude. The Amatrice’s old town centre was literally razed to the ground, accounting for more than 3/4 of the 300 victims of the earthquake. Different factors could have contributed to such a large amount of fatalities, including building vulnerability and local amplification of ground motion. Therefore, in view of the planning of the post-earthquake rebuilding of the town, it is of the utmost importance to clarify the reasons of this disaster. One of the questions that need to be investigated is the role of site response in aggravating the earthquake damage. A first contribution to shed some light on this aspect can derive from the study of ambient noise, since the employment of single station measurements using a lightweight, compact sensor allows a quick acquisition of data even within the context of an area that has to cope with the problems of emergency management. Hence, a series of campaigns of ambient noise acquisition were arranged, using a set of 3 tromographs “Tromino”. The data were acquired keeping one of the sensor at a “reference” station, where ambient noise was recorded continuously, while the other two tromographs were used for measurement sessions of 30-46 minutes at different sites. In this way, it is possible to check if differences in ambient noise properties acquired by different stations at different times reflect spatial changes in site-specific characteristics of soil dynamic response rather than temporal changes of environmental conditions generating ground vibrations. Furthermore, this configuration of data acquisition offers the possibility of additional types of data processing (e.g. through the analysis of correlation between simultaneous recordings to derive constraints for subsoil velocity models). Noise recordings were acquired on three different dates (October and December 2016, April 2017) at sites located within the most damaged area (known as the “red zone”) and outside it. Ambient noise data acquired in this area were processed following the traditional Nakamura’s approach (Nakamura, 1989) and the one recently proposed by one of us (Del Gaudio, 2017). The traditional Nakamura’s approach calculates the average spectral ratios HVNR between horizontal and vertical component of noise recording. The new approach, through instantaneous polarization analysis, identifies packets of Rayleigh waves within the noise recording and determines their ellipticity (ratio between horizontal H and vertical V component of elliptical particle motion), together with the azimuth of the vertical plane containing the elliptical motion. This method can provide thousands of estimates of H/V ratios from instantaneous polarization properties of ground motion (hereafter named HVIP), isolating the properties of portions of noise recordings where Rayleigh waves are dominant and then inferring curves of Rayleigh wave ellipticity HVIP as function of frequency. Such curves can be related to site response properties like frequency resonance, directivity of site response and impedance contrast between surface layer and bedrock. Our initial results show that, at different sites of Amatrice, both data processing methods consistently pointed out more or less pronounced maxima of H/V ratios at similar frequencies, comprised between 2 and 4 Hz. The frequency upper bound was observed at station Ama1, at the NW limit of the red zone, the lower bound at Ama8, at the SE limit of the same zone, and an intermediate value at the “reference” station Ama2 located about 300 m away from the red zone (Figs. 1-2). The measurements carried out at different dates provided similar results at Ama2, apart from a slight rotation of the H/V maximum direction, whereas evident differences were found in H/V peak amplitude and in directivity at stations within the red zone. Comparatively, the differences appear stronger for the HVNR measurements than for the HVIP ones. This confirms what we had observed in previous tests i.e., within noise recording, the superimposition of different types of waves (Love and body waves) over Rayleigh waves enhances the variability of the results from HVNR calculations, depending on the efficiency with which certain wave types are excited under different environmental conditions. However, the HVIP-derived results show that, even isolating the contribution of Rayleigh waves to the H/V ratios, a certain variability is observed. Our previous study (Del Gaudio et al., 2014) hypothesized that variations could derive from changes of the Poisson ratio of surface layers, related to seasonal changes of water content. The verification of such hypothesis, however, will require further measurement repetitions. At station Ama1, main differences from October 2016 to April 2017 measurements concern the enlargement of the most amplified frequency band and the passage from a more directional character of the H/V peak (oriented in NNW direction at October 2016) to an almost isotropy of the H/V maximum on April 2017. This suggests that the directivity observed in the first campaign at Ama1 might not reflect a property of the site response, being conditioned by changing environmental conditions (e.g., the presence and spatial distribution of sources of more or less polarized noise around the measurement station). More striking appears the variation at the station Ama11, located at the center of the old town, near the clock tower. The results obtained on April 2017 showed a strong generalized reduction of H/V ratios, so that on the basis of this measurement alone, one would totally exclude the presence of site amplification. However, it would appear that evidences of amplification are less pronounced at the sites in the central part of the red zone with respect to those near its limits. At present we do not have a straightforward explanation for these spatial differences in amplification. Nevertheless, in general, the amplification registered at Amatrice can be linked to the local geologic setting and in particular to the presence of the impedance contrast between few tens of meters thick surficial Quaternary alluvial deposits and the underlying flysch-like bedrock. We can speculate that the above-mentioned spatial variations in the amplification could result from local changes in the bedrock lithology, with more rocky (sandstones) flysch sites producing greater impedance contrast than more marly-argillaceous sites. Overall, the extension and repetition of noise measurements are necessary to achieve a better comprehension of the properties of site response in different part of the Amatrice town and to pin point possible correlations with the damage distribution. These additional measurements should help us to better distinguish noise characteristics controlled by site condition peculiarities from those depending on changes of noise sources.
Site response characteristics at Amatrice from ambient noise analysis
Vincenzo Del Gaudio;Antonio Moretti;
2017-01-01
Abstract
The 6.0 Mw earthquake that hit central Italy on 24 August 2016 caused at Amatrice damage of unprecedented severity for an earthquake of such a magnitude. The Amatrice’s old town centre was literally razed to the ground, accounting for more than 3/4 of the 300 victims of the earthquake. Different factors could have contributed to such a large amount of fatalities, including building vulnerability and local amplification of ground motion. Therefore, in view of the planning of the post-earthquake rebuilding of the town, it is of the utmost importance to clarify the reasons of this disaster. One of the questions that need to be investigated is the role of site response in aggravating the earthquake damage. A first contribution to shed some light on this aspect can derive from the study of ambient noise, since the employment of single station measurements using a lightweight, compact sensor allows a quick acquisition of data even within the context of an area that has to cope with the problems of emergency management. Hence, a series of campaigns of ambient noise acquisition were arranged, using a set of 3 tromographs “Tromino”. The data were acquired keeping one of the sensor at a “reference” station, where ambient noise was recorded continuously, while the other two tromographs were used for measurement sessions of 30-46 minutes at different sites. In this way, it is possible to check if differences in ambient noise properties acquired by different stations at different times reflect spatial changes in site-specific characteristics of soil dynamic response rather than temporal changes of environmental conditions generating ground vibrations. Furthermore, this configuration of data acquisition offers the possibility of additional types of data processing (e.g. through the analysis of correlation between simultaneous recordings to derive constraints for subsoil velocity models). Noise recordings were acquired on three different dates (October and December 2016, April 2017) at sites located within the most damaged area (known as the “red zone”) and outside it. Ambient noise data acquired in this area were processed following the traditional Nakamura’s approach (Nakamura, 1989) and the one recently proposed by one of us (Del Gaudio, 2017). The traditional Nakamura’s approach calculates the average spectral ratios HVNR between horizontal and vertical component of noise recording. The new approach, through instantaneous polarization analysis, identifies packets of Rayleigh waves within the noise recording and determines their ellipticity (ratio between horizontal H and vertical V component of elliptical particle motion), together with the azimuth of the vertical plane containing the elliptical motion. This method can provide thousands of estimates of H/V ratios from instantaneous polarization properties of ground motion (hereafter named HVIP), isolating the properties of portions of noise recordings where Rayleigh waves are dominant and then inferring curves of Rayleigh wave ellipticity HVIP as function of frequency. Such curves can be related to site response properties like frequency resonance, directivity of site response and impedance contrast between surface layer and bedrock. Our initial results show that, at different sites of Amatrice, both data processing methods consistently pointed out more or less pronounced maxima of H/V ratios at similar frequencies, comprised between 2 and 4 Hz. The frequency upper bound was observed at station Ama1, at the NW limit of the red zone, the lower bound at Ama8, at the SE limit of the same zone, and an intermediate value at the “reference” station Ama2 located about 300 m away from the red zone (Figs. 1-2). The measurements carried out at different dates provided similar results at Ama2, apart from a slight rotation of the H/V maximum direction, whereas evident differences were found in H/V peak amplitude and in directivity at stations within the red zone. Comparatively, the differences appear stronger for the HVNR measurements than for the HVIP ones. This confirms what we had observed in previous tests i.e., within noise recording, the superimposition of different types of waves (Love and body waves) over Rayleigh waves enhances the variability of the results from HVNR calculations, depending on the efficiency with which certain wave types are excited under different environmental conditions. However, the HVIP-derived results show that, even isolating the contribution of Rayleigh waves to the H/V ratios, a certain variability is observed. Our previous study (Del Gaudio et al., 2014) hypothesized that variations could derive from changes of the Poisson ratio of surface layers, related to seasonal changes of water content. The verification of such hypothesis, however, will require further measurement repetitions. At station Ama1, main differences from October 2016 to April 2017 measurements concern the enlargement of the most amplified frequency band and the passage from a more directional character of the H/V peak (oriented in NNW direction at October 2016) to an almost isotropy of the H/V maximum on April 2017. This suggests that the directivity observed in the first campaign at Ama1 might not reflect a property of the site response, being conditioned by changing environmental conditions (e.g., the presence and spatial distribution of sources of more or less polarized noise around the measurement station). More striking appears the variation at the station Ama11, located at the center of the old town, near the clock tower. The results obtained on April 2017 showed a strong generalized reduction of H/V ratios, so that on the basis of this measurement alone, one would totally exclude the presence of site amplification. However, it would appear that evidences of amplification are less pronounced at the sites in the central part of the red zone with respect to those near its limits. At present we do not have a straightforward explanation for these spatial differences in amplification. Nevertheless, in general, the amplification registered at Amatrice can be linked to the local geologic setting and in particular to the presence of the impedance contrast between few tens of meters thick surficial Quaternary alluvial deposits and the underlying flysch-like bedrock. We can speculate that the above-mentioned spatial variations in the amplification could result from local changes in the bedrock lithology, with more rocky (sandstones) flysch sites producing greater impedance contrast than more marly-argillaceous sites. Overall, the extension and repetition of noise measurements are necessary to achieve a better comprehension of the properties of site response in different part of the Amatrice town and to pin point possible correlations with the damage distribution. These additional measurements should help us to better distinguish noise characteristics controlled by site condition peculiarities from those depending on changes of noise sources.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
