Earthquake-induced landslides are responsible worldwide for significant socioeconomic losses and historically have a prominent position in the list of natural hazards affecting the Iran plateau. As a step toward the development of tools for the assessment and the management of this kind of hazard at regional scale, an empirical estimator of coseismic displacements along potential sliding surfaces was obtained through a regression analysis for the Zagros region, a mountainous Iranian region subjected to earthquake-induced landslides. This estimator, based on the Newmark’s model, allows to evaluate the expected permanent displacement (named ‘‘Newmark displacement’’) induced by seismic shaking of defined energy on potential sliding surface characterized by a given critical acceleration. To produce regression models for Newmark displacement estimators, a data set was constructed for different critical acceleration values on the basis of 108 accelerometric recordings from 80 Iranian earthquakes with moment magnitudes between 3.6 and 7. The empirical estimator has a general form, proposed by Jibson (Eng Geol 91:209–218, 2007), relating Newmark displacement to Arias intensity (as parameter representing the energy of the seismic forces) and to critical acceleration (as parameter representing the dynamic shear resistance of the sliding mass). As an example of application, this relation was employed to provide a basic document for earthquake-induced landslide hazard assessment at regional scale, according to a method proposed by Del Gaudio et al. (Bull Seismol Soc Am 93:557–569, 2003), applied to the whole Iranian territory, including Zagros region. This method consists in evaluating the shear resistance required to slopes to limit the occurrence of seismically induced failures, on the basis of the Newmark’s model. The obtained results show that the exposure to landslide seismic induction is maximum in the Alborz Mountains region, where critical accelerations up to *0.1 g are required to limit the probability of seismic triggering of coherent type landslides within 10% in 50 years.

Understanding where seismically induced landslides are most likely to occur is crucial in land use planning and civil protection actions aimed at reducing property damage and loss of life in future earthquakes. For this purpose an approach proposed by Del Gaudio et al. [1] has been applied to the whole Iranian territory to provide the basis to assess location and temporal recurrence of conditions of seismic activation of slope failures, according to the Newmark's model [2]. Following this approach, occurrence probabilities for different levels of seismic shaking in a time interval of interest (50 years) were first obtained through a standard hazard estimate procedure. Then, empirical formulae in the form proposed by Jibson et al. [3] and calibrated for the main seismogenic Iranian regions were used to evaluate the slope critical acceleration (Ac)x for which a prefixed probability exists that, under seismic shakings, Newmark's displacement DN exceeds a threshold×corresponding to landslide triggering conditions. The obtained (Ac)x values represent the minimum slope resistance required to limit the probability of landslide seismic triggering within the prefixed value. A map reporting the spatial distribution of these values gives comparative indications on regional different exposure of slopes to shaking capable of inducing failures and provides a reference for hazard estimate at local scale. The obtained results show that the exposure to landslide seismic induction is maximum in the Alborz Mountains region, where critical accelerations up to ∼0.1 g are required to limit the probability of seismic triggering of coherent type landslides within 10% in 50 years.

The influence of site effects on landslide triggering during earthquakes has been inferred in several studies, but its evaluation is made difficult by the complexity of factors controlling the dynamic response of potentially unstable slopes and also by the lack of local ground motion instrumental observations. This work explores this problem and reports new findings based on an ongoing long term accelerometric monitoring conducted on a landslide-prone test area in the Apennine Mountains, Italy, where the presence of site effects enhancing seismic susceptibility of local slopes has been invoked on the basis of historic accounts of landsliding triggered at large epicentral distance. The recordings relative to low-to-moderate magnitude earthquakes showed significant amplifications affecting hillslope portions covered by thick (N5 m) colluvia and pronounced amplification maxima oriented along the local maximum slope direction on a recent deep-seated landslide. While the amplifications seem most likely linked to high impedance contrast between surface materials and underlying substratum, the causes of directivity are less clear. The case of the monitored test site together with evidence of site response directivity identified on other hillslopes, suggest that the directivity phenomena can result from a combination of topographic, lithological and structural factors that act together to re-distribute shaking energy, focusing it on site-specific directions. Thus, it is difficult to single out the critical factors controlling such phenomena and no general criterion for the identification of sites affected by directivity is proposed here. Nevertheless, the presence and orientation of site response directivity can be revealed through reconnaissance techniques by using recordings of seismic weak motion and/or ambient microtremors, and calculating azimuthal variation of shaking energy and horizontal-to-vertical ground motion spectral ratios. A comparison with the recordings obtained during the recent MW=6.3 earthquake that hit the Abruzzo region in April 2009 demonstrated that analysis relying on data from low energy events can furnish valid indications for slope behaviour also under stronger shaking, provided the data are well differentiated in terms of distance, azimuth and source characteristics. Furthermore, the comparative analysis of the Abruzzo earthquake recordings at a landslide and reference sites showed that directivity properties of strong shaking had been correctly anticipated using earlier weak motion observations. However, under the complex slope conditions the identification of resonance frequencies from horizontal-to-vertical spectral ratios estimated from weak-motion accelerometric recordings does not seem reliable, and better results have been obtained by velocimetric microtremor recordings.

In recent years, several workers have found numerous cases of sites characterised by significant azimuthal variation of dynamic response to seismic shaking. The causes of this phenomenon are still unclear and possibly related to combinations of geological and geomorphological factors determining a polarisation of resonance effects. To improve their comprehension, it would be desirable to extend the database of observations on this phenomenon. Thus, considering that unrevealed cases of site response directivity can be “hidden” among the sites of accelerometer networks, we developed a two-stages approach of data mining from existing strong motion database, to identify sites affected by directional amplification. The proposed procedure first calculates Arias Intensity tensor components from accelerometer recordings of each site, to determine mean directional variations of total shaking energy. Then, at the sites where a significant anisotropy appears in ground motion, azimuthal variations of HVSR values (spectral ratios between horizontal and vertical components of recordings) are analysed to confirm the occurrence of site resonance conditions. We applied this technique to a database of recordings acquired by accelerometer stations in the Iranian area. The results of this investigation pointed out some sites affected by directional resonance that appear correlated to the orientation of local tectonic lineaments, these being mostly transversal to the direction of maximum shaking. Comparing Arias Intensities observed at these sites with theoretical estimates provided by ground motion prediction equations (GMPE), the presence of significant site amplifications was confirmed. The amounts of the amplification factors appear correlated to the results of HVSR analysis, even though the pattern of dispersion of HVSR values suggests that, while high peak values of spectral ratios are indicative of strong amplifications, lower values do not necessarily imply lower amplification factors.

In relation to the assessment of earthquake-induced landslide hazard, this paper discusses general principles and describes implementation criteria for seismic hazard estimates in landslide-prone regions. These criteria were worked out during the preparation of a hazard map belonging to the official Italian geological cartography and they are proposed as guidelines for future compilation of similar maps. In the presented case study, we used a procedure for the assessment of seismic hazard impact on slope stability adopting Arias intensity Ia as seismic shaking parameter and critical acceleration ac as parameter representing slope strength to failures induced by seismic shaking. According to this procedure, after a preliminary comparison of estimated historical maximum values of Ia with values proposed in literature as landslide-triggering thresholds, a probabilistic approach, based on the Newmark’s model, is adopted: it allows to estimate the minimum critical acceleration ac required for a slope to keep under a prefixed value, the probability of failures induced by seismic shakings expected in a given time interval. In this way, one can prepare seismic hazard maps where seismic shaking is expressed in an indirect way through a parameter (the critical acceleration) representing the ‘‘strength’’ that seismic shakings mobilise in slope materials (strength demand) with a prefixed exceedance probability. This approach was applied to an area of Daunia (Apulia—southern Italy) affected by frequent landslide phenomena. The obtained results indicate that shakings with a significant slope destabilisation potential can be expected particularly in the northwestern part of the area, which is exposed to the seismic activity of Apennine tectonic structures.

A technique that analyses the horizontal-to-vertical spectral ratios (HVNR) of ambient noise recordings was tested as a tool to investigate the dynamic response of landslide prone slopes to seismic shaking. Tests were conducted in landslide areas where available accelerometer recordings allowed a comparison with amplification properties observed during earthquakes. Where accelerometer data revealed the presence of directional resonance, the HVNR technique proved to be able to recognise its presence and characteristics (orientation and frequency). The actual amplification factor does not seem to be quantifiable from the spectral ratios, whose peak amplitude values show a scattering apparently depending on changeable environmental conditions. Where site response directivity is absent, the HVNR values generally do not show a common preferential orientation of strongly directional peaks, unless, occasionally, in presence of temporary active sources of polarized noise. Thus, to avoid misinterpretations we recommend HVNR measurement repetitions at different times and under different environmental conditions.

Recent observations have shown that slope response to seismic shaking can be characterised by directional variations of a factor of 2-3 or larger, with maxima oriented along local topography features (e.g. maximum slope direction). This phenomenon appears influenced by slope material properties and has occasionally been detected on landslide-prone slopes, where a down-slope directed amplification could enhance susceptibility to seismically-induced landsliding. The exact conditions for the occurrence of directional amplification remain still unclear and the implementation of investigation techniques capable to reveal the presence of such phenomena is desirable. To this purpose we tested the applicability of a method commonly used to evaluate site resonance properties (Horizontal to Vertical Noise Ratio – HVNR or Nakamura’s method) as reconnaissance technique for the identification of site response directivity. Measurements of the azimuthal variation of H/V spectral ratios (i.e. between horizontal and vertical component) of ambient microtremors were conducted in a landslide-prone study area of central Italy where a local accelerometric network had previously provided evidence of directivity phenomena on some slopes. The test results were compared with average H/V spectral ratios obtained for low-to-moderate earthquakes recorded by the accelerometric stations. In general, noise and seismic recordings provided different amplitudes of spectral ratios at similar frequencies, likely because of differences in signal and instrument characteristics. Nevertheless, both kinds of recordings showed that at sites affected by site response directivity major H/V peaks have orientations consistent (within 20°-30°) with the direction of maximum shaking energy. Therefore, HVNR appears to be a promising technique for identifying seismic response directivity. Furthermore, in a comparative test conducted on a slope mantled in part by a deep-seated landslide we detected spectral peaks with orientations close to the maximum slope direction, whereas no evidence of directivity was found outside the slide boundaries. This indicates the influence of the landslide body on seismic response directivity.

Arias intensity is considered a shaking parameter suitable for characterizing earthquake impact on ground stability. Within the framework of a study aimed at providing tools for the assessment of hazards related with earthquake induced slope failures, Arias intensity attenuation relations were determined for the Zagros Mountains region, an active tectonic belt elongated NW-SE in the western and south-western part of Iran. The calculation of relation coefficients was based on strong-motion data of earthquakes located in the Zagros area and recorded by Iranian stations managed by BHRC (Building and Housing Research Center of Iran). Five models of attenuation relation were considered and their coefficients were estimated through a least square regression analysis. The relations obtained were then applied to a data sample different from that used for regression and the RMS (Root Mean Square) of residuals was examined in order to compare effectiveness of different relations in probabilistic estimates. Furthermore a comparison made with attenuation relations obtained *Manuscript Click here to view linked References 2 for the Alborz and central part of Iran showed significant differences possibly related to structural differences.

The complexity of factors controlling the dynamic response of marginally stable slopes and the scarcity of direct ground motion recordings acquired on landslide-prone areas make it difficult to evaluate the role of site response in seismic landslide triggering. A long term accelerometric monitoring, conducted at a tectonically and geomorphologically active site of Abruzzo in Central Italy (Caramanico Terme) has provided interesting evidence of amplification with a pronounced directional character parallel to the local slope direction on a landslide consisting of colluvial deposits overlying mudstone substratum. However, until 2008, these observations were based only on recordings of events of low to moderate magnitude. The 6.3 Mw earthquake that on 6 April 2009 hit L’Aquila, 60 km from Caramanico, allowed to test whether these previous findings hold true also for the landslide site response at higher shaking levels. The comparison of the 2009 mainshock recordings of the accelerometric station located on the landslide (CAR2) to those from two nearby stations, one sited on soft soils similar to landslide substratum (CAR1) and the other on colluvial material (CAR5), showed relative amplifications (in terms of peak horizontal acceleration PHA) very close to the average values estimated from smaller events (about 1.5 and 1.0 relatively to CAR1 and CAR5, respectively). The similarity of PHA observed on colluvium, regardless of its involvement in landsliding, confirmed that, in terms of total shaking energy, the main factor controlling amplification is the impedance contrast between colluvium and mudstone substratum. The comparison between the station on landslide and a reference station on rock (CAR4) showed a relative amplification increasing with magnitude, probably because of the stronger response of the rock site to the higher frequencies prevailing in wavetrains coming from nearby small sources. This suggests that in such cases amplification assessments based on weak motion data can lead to considerable underestimates. On the other hand, the landslide site response directivity showed similar properties for the entire range of the observed magnitudes (1.4 – 6.3). It is possible that in the near field source effects modify the ratio between directional maximum and minimum of shaking energy, without, however, significantly altering the orientation of shaking maxima. In terms of spectral properties, directivity of major peaks in horizontal to spectral ratios (HVSR) was the same at any magnitude, even though at higher magnitudes spectral ratio amplitudes tend to decrease at higher frequencies and increase at lower ones. However the comparison of HVSR with standard spectral ratios (SSR) between the station on landslide and the reference site on rock indicated that the inferences on resonance frequencies derived from single station seismic weak motion measurements (like HVSR) could be unreliable under the complex conditions of a landslide-prone slope. It appears that more reliable indications can be derived from ambient noise measurement acquired with velocimetric instruments. Thus, overall, weak motion recordings proved to provide useful information on landslide site response characteristics, especially having a dataset sufficiently differentiated in terms of azimuth location, distance, energy and source characteristics. However, the extrapolation of inferences based on recordings of small magnitude events to stronger earthquake scenarios requires some caution.

The Mw 7.9 Wenchuan earthquake of May 12th, 2008 shattered and induced failures on slopes of hills surroundings the centre of the town of Qiaozhuang, located 300 Km NE of the mainshock epicentre. This motivated the initial investigations on the possible occurrence of topographic amplification phenomena, which were conducted through a temporary accelerometer monitoring carried out between April and October 2009 on Weigan hill and Mount Dong, located SW and NE of the Qiaozhuang centre, respectively. In both cases, the amplification factors were found higher at sites closer to the hilltop and lower near the foot of the slope (Luo et al., 2014). However, the general characteristics of slope response also revealed that resonance phenomena were controlled by the local geological setting characterised by Silurian phyllites with sub-vertical schistosity (at Weigan) and by fractured Precambrian limestones (at Dong). To further investigate local slope response, ambient noise recordings, based on the calculation of horizontal to vertical spectral ratio (HVNR or Nakamura method), were also conducted on several sites in the peri-urban area, including those occupied by the accelerometer stations. The latter provided the possibility to validate the outcomes of ambient noise analysis. Data interpretation has been hampered by a contingent factor, i.e. the occurrence of environmental conditions characterised by a strong E-W oriented peak at low frequencies (below 1 Hz), and by the locally complex structural setting. For example, in the case of Weigan hill the sub-vertical schistosity appeared responsible for a considerable amplification of vertical ground motion. The former problem implies the need of a repetition of measurements under different environmental conditions to distinguish permanent, site-specific properties of noise related to slope dynamic response. By altering the H/V ratio, the presence of vertical amplification represents an obstacle for a routine data analysis and a straightforward interpretation of results. Therefore, further developments in advanced techniques of noise data analysis techniques appear necessary to facilitate the extraction of reliable information on seismic response of slopes in geologically complex conditions.

Several studies have shown that the dynamic response of landslide prone slopes to seismic shaking can play an important role in failure triggering during earthquakes. It was also demonstrated that slope seismic response is often characterised by directional resonance phenomena. Directivity can be revealed by the analysis of ambient noise recordings according to a technique known as HVNR method based on the analysis of azimuthal variation of spectral ratios between the spectral amplitude of horizontal H and vertical V component of noise recording. Directional resonance is then revealed by the presence of a preferential polarisation of H/V ratio peaks, whose frequencies correspond to resonance frequencies and whose amplitudes depend on the impedance contrast between surface material and bedrock. H/V ratio amplitudes can potentially provide information also on amplification factors. However, the relation is not straightforward depending on the nature of the waves contributing to the ambient noise. Thus, it is desirable to distinguish different kinds of noise wave packets, possibly isolating the contribution of Rayleigh waves, which appear to better reflect site response properties. To identify Rayleigh wave packets in noise recording a new approach was tested, based on a technique of analysis of instantaneous polarisation. The results are promising for the investigation of site response directional properties, particularly in the case of complex site conditions, where resonance can be characterised by multiple anisotropic peaks. In our preliminary tests of noise recordings carried out at a site located on a slope affected by landslides, only a small fraction of data samples (in the order of 1 %) were identified as Rayleigh type waves: this was likely due the fact that the noise recording was dominated by an overlapping of signals with different kinds of polarisation. Thus, it was possible to recognise Rayleigh polarisation only when the energy of this kind of wave was prevalent. However, from a relatively short noise recording (in the order of 30-45 minutes) one can obtain a high number (in the order of thousands) of estimates of H/V amplitude and azimuth, providing a robust statistics to recognise ground vibration properties reflecting site response. The tests on sites where directional resonance properties had been verified through the analysis of seismic event recordings, showed that more coherent observations can be obtained for H/V ratios and directivity estimates by selecting Rayleigh type data samples, rather than analysing the entire data set or SH-type wave packets.This offers the possibility of reducing the uncertainties in data interpretation related to the influence of the nature of the noise wavefield.

Recent experimental studies have pointed out that susceptibility of slopes to seismically induced failures can be significantly increased by site conditions determining phenomena of directional resonance along potential sliding directions. Evidence of such situations were obtained from long-term accelerometric monitoring conducted in central Italy (Caramanico Terme), and it has been suggested that similar phenomena may have influenced also some of the giant landslides triggered in Taiwan by the 1999 Chi-Chi earthquake. Since factors controlling directional resonance are still unclear, it is of great interest to develop reconnaissance techniques capable to point out the occurrence of conditions of site response directivity. One promising technique is based on the analysis of the azimuthal variations of the spectral ratios between horizontal and vertical component of noise recordings (HVNR), derived from microtremors measured with portable velocimeters. At this regard an extension of observations to low frequency spectral ratios, even below 1 Hz, is desirable, especially in the case of very large landslides whose stability conditions under seismic shaking could be more sensitive to the effect of larger wavelengths. For this reason we tested the employment of a portable broad-band sensor for HVNR measurements on landslide-prone slopes. First, measurements previously carried out at Caramanico with different seismometric sensors were repeated for comparative purposes. Then, data acquisition was extended to other sites in the same area and in two areas of Taiwan, on the giant landslides of Tsaoling and Jiufengershan. It is recognized that the analysis of noise recording on a broader frequency bandwidth poses some supplementary difficulties for the presence of strong low frequency signals that vary with atmospheric and sea conditions. However, the measurements obtained at Caramanico showed that, where clear site response directivity had been revealed by previous observations, the results of HVNR analysis derived from broad-band measurements were consistent with those derived by other type of sensors. At other Caramanico sites, HVNR analysis of the broad-band data provided uncertain results, possibly as effect of variable conditions of data acquisition that might obscure weaker site response effects. In some cases the repeatability of the results was a problem, possibly due to temporary action of a local source of strong polarised noise. The measurements obtained in Taiwan in the two giant landslide areas, both characterised by mass movements occurred along bedding surface, revealed the presence of abnormally high values of HVNR for sites on slip surfaces, with a very strong directional variation through a large frequency band and with pronounced maxima at frequency less than 1-2 Hz, i.e. at frequencies influenced by signals of atmospheric and marine origin. Although the recordings on nearby sites on the flanks or the crowns of the landslides did not reveal similar properties, the representativeness and exact significance of the HVNR data acquired on the slip surfaces need to be more deeply investigated in order to exclude the effects of anomalous acquisition conditions. More in general, contemporaneous measurements of noise with two different types of instruments at different times and varying environmental conditions should help to clarify to what extent the reconnaissance type of noise measurements by broad-band instruments could be useful for detecting directivity in seismic site response

The improvement in the Civil Protection system capability of mitigating the impact of future large magnitude earthquakes requires an upgrading of methods aimed at providing realistic scenarios of seismic effects. One of the factors that contributes the most to the uncertainty of the previsional assessment of earthquake consequences is the dynamic response of ground to seismic shaking. This is especially relevant in complex geological-geomorphological settings, which can determine dramatic variations of shaking energy and ground failure effects, even within very short distances. The most straightforward approach to the investigation of ground dynamic response consists of the direct recording of seismic ground motion at sites suspected to be liable to amplification phenomena. However, while it is not difficult to collect weak motion observations with relatively short term monitoring, their scaling to strong shaking conditions can provide unreliable results. On the other hand, much time is usually necessary to acquire strong motion data. This and the cost of instruments needed to provide suitable resolution level for site response investigations in geologically heterogeneous areas makes long-term widespread monitoring impractical. These problems promoted the experimentation of alternative approaches based on the analysis of persistent ground vibrations generated by non-seismic sources (wind, sea waves, traffic, etc.), commonly defined as “ambient noise”. The noise signals can be recorded through short time (less than 1 hr) acquisition sessions conducted with lightweight portable instruments, even at sites of difficult accessibility. The outcomes of ambient noise analysis suffer from uncertainties comparable to those of weak motion data interpretation. Nevertheless, noise analysis can benefit from the possibility of an extensive application, also to sites recently hit by strong earthquakes. Such events offer comparative data (e.g. seismic recordings) that can be used to refine the information derivable from noise signals. The most popular technique of ambient noise analysis for site response investigation is the Nakamura (1989) method, also named HVNR (horizontal-to-vertical noise ratios). It relies on the analysis of ratios H/V between spectral amplitudes of horizontal (H) and vertical (V) components of noise recordings, subdivided into series of 20-30 second long time windows. Averaging spectral ratios over several tens of such time windows, the occurrence of amplification conditions can be recognized from the presence of pronounced peaks at site specific frequencies. These were found to correspond to the resonance frequencies deriving from conditions of seismic wave constructive interference, controlled by geometrical and physical properties of surficial geological bodies. Although originally devised to investigate site response of sites with flat horizontal layering, where amplification is caused by wave trapping inside a soft surface layer overlying a stiffer substratum, the Nakamura’s method proved able to reveal resonance conditions also under more complex site conditions (e.g., slopes/topographic relieves with lateral variations of subsoil materials and their mechanical properties). In such conditions, an analysis of azimuthal variation of spectral ratios H/V can show directional variations consistent with those observed in ground motion amplification. The improvement in the Civil Protection system capability of mitigating the impact of future large magnitude earthquakes requires an upgrading of methods aimed at providing realistic scenarios of seismic effects. One of the factors that contributes the most to the uncertainty of the previsional assessment of earthquake consequences is the dynamic response of ground to seismic shaking. This is especially relevant in complex geological-geomorphological settings, which can determine dramatic variations of shaking energy and

This work reports the results of an accelerometer monitoring aimed at revealing the seismic response of hillslopes in the area of Qiaozhuan town (Qingchuan county), located near the north-eastern end of the fault ruptured during the 7.9 Mw 2008 Wenchuan earthquake (Sichuan Province). Serious damage and slope failures were induced by this earthquake in the town center and on the hills in the peri-urban zone. This suggested the possible occurrence of amplification phenomena. Five recording stations were emplaced at two topographic reliefs to investigate their response to ground motion during the last part of the Wenchuan seismic sequence. About 50 aftershocks were recorded, whose magnitude (ML) varied between 1.2~5.5 and epicentral distance ranged from a few to 90 km. The recordings provided evidence of directional amplification. This phenomenon was investigated by analyzing polar diagrams of normalized Arias intensity (Ia) and horizontal to vertical spectral ratios (HVSR) to find, respectively, polarization azimuth and resonance frequencies. Evidences of anisotropic dynamic response were found at both investigated topographic reliefs, but with ground shaking maxima differently oriented with respect to the relief elongation: in one case transversal and in the other case sub-parallel. No clear preferential direction of maximum site response was observed at a site in the valley. Furthermore evidence of resonance was derived from the calculation of spectral ratio between the sites on the slope and at the foot of the hills. The resonance was more pronounced at higher elevations, which suggested a role of topographic amplification. Resonance frequencies were lower (3-5 Hz) on the hill consisting of sub-vertically layered phyllites, and higher (up to 7 Hz) on the other, larger hill made mainly of limestones, while an opposite relation between resonance frequency and hill size could be expected from a purely topographic effect. This and the amplification factor larger than 2 suggest that in addition to topographic effects also local geology played a significant role in differentiating the site response.

We report on the seismic response of slopes in the area of Qiaozhuang town (Qingchuan county), located 250 Km N-E of the epicenter of the 2008Wenchuan earthquake (Sichuan Province). The earthquake caused significant damages on the slopes surrounding the town, including the Weigan hill, which was affected by diffuse opening of cracks. This suggested the possible occurrence of topographic amplification phenomena and motivated a subsequent in situ accelerometer monitoring. Recording stations were emplaced at the top and the toe of the Weigan hill, as well as at other sites located on slopes in the N-E periphery of the town. About 100 aftershocks of the Wenchuan sequence were recorded, whose magnitude varied between 1.2 _ 5.5 and epicentral distance from a few to 103 km. A preliminary analysis of the the Weigan hill recordings provided evidence of the presence of directional variation of ground vibration possibly related to directional resonance. This phenomenon was first investigated by analyzing polar diagrams of normalized Arias intensity (Ia) and horizontal to vertical spectral ratio (HVSR) to find, respectively, polarization azimuth and resonance frequencies. The most pronounced Ia directivity was observed at a site near the hilltop, where ground motion maxima were found persistently orientated around N-S direction. Furthermore evidence of significant amplification was derived from HVSR data and from the comparison to other accelerometer monitoring points. Then, we conducted ambient noise measurements aimed at examining the azimuthal variation of the horizontal to vertical spectral ratios of noise recordings (HVNR) at the accelerometer station sites. This provided the possibility to test the reliability of site response directivity assessment inferred from ambient noise analysis. Noise measurements were carried out with two tromographs, using one of them as a continuously recording reference, while the second tromograph was moved through other sites for recording sessions of 46 minutes. The analysis of data showed the ubiquitous presence of a strong E-W oriented peak at low frequencies (below 1 Hz), likely related to the influence of a remote source of noise, as well as the presence of locally varying peaks. In particular, a preferential orientation of maximum ground vibration in N-S direction was found at the site near the Weigan hilltop, even though the spectral ratio amplitudes were not very high. This suggests the need of more advanced analysis to filter disturbing signals. In summary, the indications of directivity phenomena obtained through ambient noise data analysis were generally consistent with the findings based on the accelerometer data.

In 2004 an earthquake of magnitude 6.8 hit the central part of the Niigata Prefecture in Japan, causing high damage and thousands of landslides. Two among the most catastrophic of these landslides occurred at the localities of Terano and Higashi Takezawa, causing damming of the Imo river and severe flooding. In both cases the estimated volume of the displaced material exceeded one million cubic meters. The lithologies involved in failures were sandy silt, silty sand, sandstone, siltstone and mudstones. Following recent studies on phenomena of directional resonance characterising the dynamic response of some slopes prone to landsliding, in situ recordings of seismic noise were carried out at several sites on the slopes of the two landslides to detect possible presence of site amplification effects that may have favoured the earthquake triggering of these failures. The subsequent analysis was conducted following two procedures: i) the standard Nakamura (or HVNR) technique, based on the examination of azimuthal variations of ratios between spectra of horizontal and vertical components of noise recording; ii) a new technique that estimates the rate of recurrence, in noise recording, of signals showing coherent preferential directivity and calculates H/V spectral ratio average restricting it to peaks satisfying purposely defined significance requirements. The results of these analyses showed that, whereas the employment of the standard method provided uncertain outcomes in terms of the presence or absence of site response directivity, the new technique proved to be effective in revealing directional resonance properties. In particular, apart from an ubiquitous low frequency signal with a strong directional character, possibly related to sea wave influence, evidences of directional amplification with orientation consistent with landsliding directions (approx. west) were found on both slopes. Nevertheless, a considerable variability of resonance frequency and amplification factors was also observed, being likely related to local differences in measurement site conditions (geology and topography).

The difficulty in identifying factors controlling the dynamic response of landslide-prone slopes to seismic shaking makes desirable the development of reconnaissance techniques to reveal site resonance conditions that can favour seismically-induced slope failures. Tests were performed to derive information on the occurrence of directional resonance by analysing the azimuthal variations of the horizontal-to-vertical spectral ratios of ambient noise (HVNR) recorded by portable seismometers. In particular, data were acquired in an area of central Italy (Caramanico Terme), affected in the past by seismically-induced landslides, and in two areas of Taiwan (Tsaoling and Jiufengershan), where the 1999 Chi-Chi earthquake triggered giant landslides. The HVNR analysis demonstrated that the presence of a pronounced directional resonance can be recognised from data acquired under different ambient conditions and with different sensors. However, measurement repetitions and uncertainty assessment are fundamental to distinguish persistent features, attributable to site effects, from transient phenomena due to variable ambient conditions.

Il concetto di pericolosità sismica misura la probabilità che in un dato luogo e in un definito arco di tempo si verifichino scuotimenti sismici potenzialmente capaci di produrre effetti di danneggiamento. Le stime di pericolosità sono finalizzate a definire i livelli di scuotimento rispetto ai quali si vuole garantire un margine di sicurezza rispetto al verificarsi di danneggiamenti. Ciò è ottenuto attraverso il calcolo di livelli di scuotimento (per esempio accelerazioni del suolo) che hanno una definita ‘probabilità di eccedenza’, cioè una definita probabilità di essere superati, dove tale probabilità è fissata a rappresentare un livello di ‘rischio accettabilmente basso’. Tali livelli di scuotimento servono a definire le azioni sismiche di progetto, cioè le caratteristiche degli scuotimenti sismici rispetto ai quali occorre prendere misure di tutela per mantenere il rischio di danneggiamento entro i limiti fissati. I dati su cui si basano le stime di pericolosità finalizzate al calcolo di tali azioni sismiche di progetto sono, in Italia, principalmente costituiti da informazioni sulla sismicità storica. Attraverso un’analisi congiunta di esperti di sismologia e di esperti nell’interpretazione di fonti storiche che descrivono gli effetti di terremoti, si possono determinare i parametri essenziali di eventi sismici su lunghi intervalli di tempo, per i quali non si dispone di osservazioni strumentali. L’applicazione di questo approccio allo studio della pericolosità sismica della Capitanata ha messo in evidenza come la sua pericolosità, pur essendo inferiore a quella delle vicine regioni appenniniche, è tutt’altro che trascurabile, dal momento che ci sono evidenze storiche di eventi sismici di elevata energia (magnitudo pari o superiore a 6.0) in diverse occasioni, con effetti devastanti e vittime anche dell’ordine delle migliaia. Il fatto che tali eventi abbiano una ricorrenza temporale meno frequente rispetto alle aree appenniniche determina maggiori incertezze nelle stime statistiche della pericolosità, ma un supporto al miglioramento di queste valutazioni può derivare dall’integrazione dei dati storici con le osservazioni strumentali di eventi di bassa energia. I dati sui piccoli terremoti consentono infatti di meglio vincolare le stime dei tempi medi di ricorrenza anche dei terremoti più grandi. Alcuni aspetti critici di tali stime riguardano l’identificazione delle sorgenti dei maggiori terremoti del passato, che potenzialmente potrebbero riattivarsi in futuro, e la valutazione di come condizioni geologiche locali di amplificazione delle onde sismiche possono rendere pericolosi anche i risentimenti di terremoti originati in aree contigue, dove forti terremoti avvengono più di frequente. Su queste tematiche è importante investire risorse per migliorare le conoscenze sulla pericolosità della Capitanata, particolarmente con riferimento a misure di protezione sismica. Tali misure vanno pensate non solo rispetto dell’edificato comune ma anche per il patrimonio storico, artistico, monumentale, per il quale le azioni sismiche di progetto vanno valutate con specifico riferimento a un più lungo orizzonte temporale, entro il quale si ha interesse a salvaguardare il patrimonio culturale del territorio.

Uncertainty in the quantification of earthquake loading poses one of the major difficulties in local scale seismic landslide susceptibility assessments. This problem can be exacerbated for slope settings that are likely to produce considerable amplifications of seismic shaking. We address this issue by examining the case of a historic landslide triggered by the 1627 Apulian (southern Italy) earthquake (epicentral intensity X on the MCS scale) in the peri-urban area of Caramanico (central Italy), distant ~120 km from the epicenter. The failure caused a large downslope displacement and destroyed several buildings. The slope seismic response is assessed using data from long-term accelerometer monitoring of the hillslope and from recent ambient noise measurements. This provided evidence of significant directional amplifications, e.g., by a factor of approximately 4 and 20, respectively in terms of peak horizontal acceleration and Arias Intensity during the 2009 Mw 6.3 L’Aquila earthquake that occurred 60 km NW of Caramanico. Then taking into account the site amplification, permanent displacements are calculated by applying a rigorous Newmark approach. This study shows that historical information on landslides triggered at apparently anomalously large distances from an earthquake epicentre can help to identify hillslopes influenced by site effects and that reconnaissancetype measurements of ambient noise can be useful to reveal directional amplifications. The importance of accurate assessments of other relevant input parameters (e.g., material properties, slip surface geometries, groundwater conditions) used in seismic slope modeling is also recognized.

Slope stability analysis in seismically active areas must take into consideration the influence of site response during earthquakes. While widespread accelerome-ter monitoring of slopes appears impractical, it is possible to derive information relevant for stability conditions from passive seismic techniques based on the ac-quisition and analysis of ambient noise generated by natural and anthropic sources. A useful technique is Nakamura’s method, which consists in analysing ratios between horizontal (H) and vertical (V) component spectra of noise record-ings. An analysis of azimuthal variation of H/V can reveal directional resonance phenomena affecting slope areas, providing the frequency and direction of max-imum ground motion amplification. Determination of site response properties from seismic noise mainly depends on the identification of polarisation direction and ellipticity of Rayleigh waves. However, in ambient noise records only part of acquired signals may show coherent characteristics referable to Rayleigh waves. Thus, it is useful to develop techniques that allow selecting, within noise record-ings, wave trains that can be reliably identified as Rayleigh waves. A new promis-ing technique is based on the identification of instantaneous polarisation proper-ties from analytical signal transformation. The study of slope dynamic response to shaking can also benefit from the analysis of cross-correlation among simulta-neous noise recordings. This type of analysis provides dispersion curves for Ray-leigh waves, which can then be used to constrain S-wave velocity models and, consequently, to infer mechanical properties of slope materials.

La principale linea di difesa adottata attualmente per la mitigazione degli effetti dei terremoti si basa su un approccio preventivo che dispone, nella progettazione di opere ingegneristiche e nella pianificazione territoriale, misure atte a rendere il territorio in grado di sopportare l’impatto di eventi sismici futuri, contenendone i danni. Ciò richiede una stima della pericolosità sismica del territorio, cioè dei livelli di scuotimento attesi, rispetto ai quali occorre predisporre azioni di protezione. Il calcolo della pericolosità è effettuato su base statistica analizzando le serie storiche dei terremoti, ma l’affidabilità di queste stime dipende fortemente dalla ricchezza di informazioni disponibili sugli eventi sismici del passato. Ciò rende più incerte questo tipo di stime per regioni, come quella pugliese, dove, in presenza di lunghi tempi medi di ritorno dei terremoti ad elevata energia, la base di dati è più povera. In Puglia, infatti, eventi che hanno prodotto un numero di vittime nell’ordine delle centinaia o delle migliaia, si sono verificati in più occasioni, ma, a parte un caso isolato (Ascoli Satriano, 1361), tutte concentrate in un arco di circa 120 anni dal 1627 al 1743. Quasi tutti i maggiori terremoti hanno interessato il nord della regione, ma l’ultimo evento di questa serie ha colpito il Salento, pur essendo la sorgente localizzata al largo delle sue coste, con un pesante bilancio di vittime prodotto da fenomeni di amplificazione locale degli scuotimenti, legati alla particolare geologia di questo territorio. E’ quindi essenziale non dimenticare che tutto il territorio pugliese non può essere considerato al sicuro dagli effetti dei terremoti e che misure di prevenzione dal danno sismico vanno comunque previste, con il supporto di indagini finalizzate a migliorare la conoscenza del territorio ed a colmare le lacune presenti nella base di dati attualmente disponibile. Coltivare la memoria storica dei terremoti del passato è perciò essenziale per non indurre atteggiamenti di sottovalutazione e trascuratezza rispetto alla necessità di una più attenta valutazione del rischio sismico presente sul territorio e per promuovere, nella progettazione degli edifici e nella pianificazione territoriale, l’adozione di cautele che ne tengano conto.

Defining the possible scenario of earthquake-induced landslides, Arias Intensity is frequently used as shaking parameter, being considered the most suitable for characterizing earthquake impact, while Newmark's sliding-block model is widely used to predict the performance of natural slopes during earthquake shaking. In the present study we aim at providing tools for the assessment of the hazard related to earthquake-induced landslides at regional scale, by means of new empirical equations for the prediction of Arias Intensity along with an empirical estimator of coseismic landslide displacements based on Newmark’s model. The regression data, consisting of 205 strong motion recordings relative to 98 earthquakes, were subdivided into a training dataset, used to calculate equation parameters, and a validation dataset, used to compare the prediction performance among different possible functional forms and with equations derived from previous studies carried out for other regions using global and/or regional datasets. Equations predicting Arias Intensities expected in Greece at known distances from seismic sources of defined magnitude proved to provide more accurate estimates if site condition and focal mechanism influence can be taken into account. Concerning the empirical estimator of Newmark displacements, we conducted rigorous Newmark analysis on 267 one-component records yielding a dataset containing 507 Newmark displacements, with the aim of developing a regression equation that is more suitable and effective for the seismotectonic environment of Greece and could be used for regional-scale seismic landslide hazard mapping. The regression analysis showed a noticeable higher goodness of fit of the proposed relations compared to formulas derived from worldwide data, suggesting a significant improvement of the empirical relation effectiveness from the use of a regionally-specific strong-motion dataset.

Long term accelerometric monitoring of a tectonically and geomorphologically active site in Central Italy (Caramanico Terme) is used to study the dynamic response of slopes to seismic shaking. By combining the 2009 Mw 6.3 L’Aquila earthquake data with results based on lower magnitude events recorded earlier, it is shown that a valuable assessment of site response properties can be obtained by averaging weak motion data, provided the latter are sufficiently dif-ferentiated in terms of azimuth location, distance, energy and source characteristics. The case of directional amplification affecting a slope prone to deep-seated landsliding is described to offer some practical indications on how to estimate the presence and significance of site effects on slopes susceptible to seismic failures. For the monitored slope we estimate that amplifications of shaking energy, expressed through Arias intensity, vary on average from 2 to 16, if calculated relatively to sites on soft soil or rock, respectively.

The assessment of the influence of slope dynamic response on earthquake landslide triggering is hampered by the scarcity of accelerometric data acquired on slopes prone to seismic failures. Previous studies showed that the spectral analysis of 3-component recordings of seismic noise can represent a promising tool for the characterisation of slope dynamic response. In this case the main method of analysis is based on the calculation of horizontal-to-vertical spectral ratios of noise recordings (HVNR). However, this method was originally designed to investigate properties of litostratigraphic amplification of sites with simple setting as approximated by 1D layering; thus its application to hillslope sites affected by 2D-3D and anisotropic amplification effects requires further tests to verify which properties of seismic response can be reliably obtained. For a fruitful interpretation of seismic noise recordings the main problems that needs to be addressed are i) the extension of analysis to frequencies below 1 Hz and ii) the temporal consistency and spatial resolution of noise recordings. The extension of analysis from microtremor frequencies (>1 Hz), mainly of anthropic origin, to microseismic frequencies (< 1Hz) of natural origin is desirable, in that lower frequencies can be more relevant for triggering of large landslides. However the reliability of the results of low frequency noise analysis strongly depends on the adequate response of portable instruments at such frequencies and on the influence of changing environmental conditions. Regarding the problem of spatial-temporal variability of noise signal, we stress that directional properties of site response can be masked by polarisation properties of transient noise sources and that the spatial scale of geological and morphological features causing directivity is still unclear. Thus, the repeatability of noise analysis results, as well as their spatial resolution and representativeness need to be verified through comparisons of noise recordings suitably arranged in space and time. With this motivation a series of tests of seismic noise recordings and analyses are in progress on landslide-prone slopes at Caramanico Terme, in central Italy, where the results of noise analysis can be compared with those derived from a long term accelerometric monitoring. In particular repeated noise recordings are carried out both at the same site and at different sites using simultaneously seismic sensors having different frequency response. The first results indicate that directional properties of site response appear to characterise landslide prone slopes also at frequencies lower than 1 Hz, even though in this case ad hoc methods of analysis need to be used to reveal such effects. Furthermore, the tests of seismic noise recordings conducted at different times and with different sensors show that ground motion anisotropy due to site response directivity can be discriminated from that related to noise source polarisation by few repetitions of noise data acquisitions of proper duration.

Earthquake induced slope failures are responsible for a significant amount of life loss and damage, and their effective mitigation requires further advancements in our comprehension of slope behaviour under seismic shaking. One source of uncertainty in seismic landslide susceptibility assessment is the phenomenon of enhanced amplification of ground motion along down slope directions. This implies a strength demand beyond that estimated by standard slope stability analysis. An extensive accelerometer monitoring of slope dynamic response in areas exposed to seismic landslide hazard is unfeasible. An alternative approach can take advantage of recent development of reconnaissance techniques based on the analysis of ambient noise recorded by portable instruments. Themost popular technique, known as Nakamura or HVNRmethod, consists in analysing H/V spectral ratios between Horizontal and Vertical components of Noise Recording, and allows the recognition of site resonance frequencies. The application of HVNR to complex site conditions typical of marginally stable slopes is often difficult and requires the development of “ad hoc” procedures both for acquisition and analysis of noise recording. Tests in different geologic and geomorphic settings show that an analysis of azimuthal variation of spectral ratios can reveal the presence and orientation of directional resonance, whereas the recognition of main resonance frequencies requires a proper selection of signals to be analysed. Efforts to evaluate amplification factors currently rely on numerical simulations, which in turn require S-wave velocity of slope materials. Ambient noise analysis in terms of velocitymodels can contribute through the inversion of H/V spectral ratios and surface wave velocity dispersion curves derived from the processing of multiple simultaneous noise recordings. However these applications require a correct identification of the nature of surface waves present in the noise recording.