RICERCA ITALIANA     

INDUCED SEISMIC HAZARD: ANALYSIS, MODELLING AND PREDICTIVE SCENARIOS OF EARTHQUAKE TRIGGERED LANDSLIDES

Università degli Studi di Roma "La Sapienza"

Abstract

The recent seismic events in South-East Asia demonstrated that "collateral damage"; may significantly exceed economic and social losses directly related to seismic shaking. Landslides rank among the main categories of effects due to the release of seismic energy. Seismically-induced landslide hazard and the presence of exposed elements and their vulnerability produce risks to be mitigated and managed. Therefore, a comprehensive approach to seismic hazard assessment in a given area requires a thorough study of the mechanisms that trigger or reactivate seismically induced landslides. Landslide bodies existing prior to seismic shaking may have not yet reached rupture conditions s.s.. Nonetheless, they may amplify seismic waves, alter seismic shaking and cause damage to the built heritage.
In view of the above, the prime objective of this research program is to develop scenarios of hazards and risks arising from mobilisation of landslides due to seismic stress and from major seismic amplification processes occurring inside the landslide mass.
To this end, studies will be conducted in geological and geomorphological settings that are typical of wide areas of the Apennine range. Recent earthquakes in these areas (e.g. Umbria-Marche, Molise and Sicilia) proved to have a considerable seismic potential in terms of frequency of occurrence and released energy.
In this framework, the contributions of the Research Units engaged in the project will focus on collection of adequate data in the various fields and on identification of validating criteria:
1) earthquake-slope interaction models based on case histories;
2) seismically induced landslide risk scenarios based on regional-scale analyses;
3) based on the activities mentioned points 1) and 2) above, implementation of assessment models applicable to other geological settings, guidelines and operational procedures for risk mitigation and management.
To achieve the above goals in the study of case histories, the research program will consist of the following stages:
- building of preliminary engineering-geological models for case studies specially selected for this research program (Caramanico Terme-CH, Salcito-CB, Cerda-PA);
- performance of laboratory analyses for characterising the static and dynamic behaviour of soils that are involved in seismically induced landslides or in landsliding areas capable of amplifying seismic inputs;
- performance of in-situ engineering-geology studies for identifying geometries, kinematic processes, on-site physico-mechanical properties, as well as boundary conditions of landsliding areas;
- performance of analyses of local seismic response and definition of landslide body geometries and amplification effects;
building of reference earthquake-slope interaction models to be validated via numerical models analysing amplification conditions and landsliding mechanisms.
Based on regional-scale analyses, seismic hazard analyses will be concurrently conducted. Finally, the combination of the results from these analyses with the parameters obtained from final earthquake-slope interaction models will lead to the development of seismically induced landslide risk scenarios for specified areas of the built environment. <<<

Principal Investigator

Gabriele SCARASCIA MUGNOZZA Università degli Studi di ROMA "La Sapienza"

Research Objectives

The main objective of the Research Program is the development of scenarios of hazards and risks arising from earthquake-induced landslide mobilisation, as well as from major phenomena of seismic amplification in existing landslide bodies.
Of considerable interest will be the study of earthquake-induced landslides on slopes that consist of texturally complex clays, in view of: their dissemination in wide areas of Italy; their involvement in more or less recent earthquakes (Irpinia 1980; Umbria-Marche, 1977; Molise, 2002; Sicilia, 2002); and presence of growing number of structures and infrastructures over the country and of new residential and industrial settlements.
To attain the objective, different investigation channels will be activated. These channels, although being part of a multi-disciplinary approach, will operate synchronously throughout the research activity.
Given the ultimate aim of this Research Program, each investigation channel will reach intermediate or partial goals; however, such goals will yield aspects of considerable interest and immediate benefits in terms of applications.
The partial goals, to be set upon research planning, will identify the milestones of the Research Program:
1. through specific laboratory investigations, assessment of monotonic and cyclical behaviour of lithotypes with dominantly clayey component and complex texture (e.g. “argille varicolori”) and, namely, of their stiffness and dampening parameters, as well as assessment of the impact of their meso-structural characters on linear and volumetric threshold deformation values;
2. definition of in-situ dynamic properties via in-hole geophysical investigations and comparison with results from laboratory tests on the same texturally complex clayey lithotypes with a view to deriving correlations;
3. identification and quantification of the impact of particular topographic conditions (i.e. tectonic unconformities) on the seismic response of landslide areas (via in-situ seismometric recordings) in terms of both spectral amplification and shaking energy polarisation;
4. reconstruction of mechanisms and kinematic models typical of seismically induced landslides in structurally complex clayey lithotypes under co-seismic and post-seismic conditions;
5. building of non-linear soil behaviour constitutive models for numerical analyses via finite-difference software codes, with a view to fine-tuning the simulation of large-deformation effects;
6. development of seismically-induced landslide risk scenarios in case-study areas (to be defined jointly by the Research Units participating in the project) with deterministic and probabilistic methods, by assessing damage to local structures and infrastructures (previously censused) vs. their vulnerability functions and intensity of expected landslides.
To achieve goal no. 1, Operational Units 1 and 2 will conduct laboratory investigations in the static field to determine the influence of circulating fluids and tests in the dynamic field via a Double Specimen Direct Simple Shear Device (DSDSS) on specially collected samples.
To achieve goal no. 2, Operational Unit 1 will select the geophysical techniques most suitable for investigating the lithotypes involved in landslides; these lithotypes will be selected in order to determine their geometry and characterise their stiffness properties at low levels of deformation via active seismic prospecting.
To achieve goal 3, operational units 1 and 4 will put in place temporary seismometric arrays recording ambient noise or seismic events of low magnitude, in order to determine local seismic response via analyses of time- and frequency-domain signals and techniques based on ratios/relations between components and/or between components and a reference site.
To achieve goal no. 4, all Operational Units will fed the results from their specific site and laboratory investigations into a concise picture, herein referred to as reference earthquake-slope interaction model.
To achieve goal no. 5, Operational Units 1 and 2 will rely on one-dimensional and two-dimensional finite-difference software codes, so as to interpret the recorded local seismic effects and build constitutive models; the latter models will be aimed at refining the non-linear behaviour models of the investigated soils and, ultimately, at justifying the large deformations related to seismically-induced landslides.
Operational Unit 3 will mostly pursue goal no. 6. Based on the control factors (check-points) identified by all other Operational Units via local investigations on specific case histories, Operational unit 3 will develop a GIS-supported process for elucidating landslide susceptibility to seismic induction and, accordingly, for describing risk scenarios.
The Operational Units will work simultaneously during the four six-month periods of the project, except for the first six-month period. Indeed, in the latter period, Operational unit 2 will not carry out the activities falling under its responsibility but it will await the end of the sampling program, which is planned to be implemented by Operational Units 1 and 4. <<<

Timescale

24 months

National and international background

Seismically-induced landslides are historically documented for some ancient earthquakes, such as the 1783 Calabria event (Sarconi, 1784) and the 1786 Kanding-Luding (China) event (Dai et al., 2005), and for recent times there is a large database related to earthquake-induced landslides (Keefer, 1984; Rodriguez et al., 1999).
Landslides are one of the most damaging collateral hazards associated with earthquakes. In fact, damage from triggered landslides and other ground failures has sometimes exceeded damage directly related to strong shaking and fault rupture (Bird & Bommer, 2004). Seismically triggered landslides damage and destroy homes and other structures, block roads, sever pipelines and other utility lifelines, and cause damming stream drainages. Predicting where and in what shaking conditions earthquakes are likely to trigger landslides is a key element in regional seismic hazard assessment.
Local civil protection Agencies need to plan their emergency activities prior the disaster occurence, thus to be prepared to promptly bring assistance to the injured population. In this perspective, one of the most useful and effective tool is the formulation of scenarios, describing how the dangerous event manifest itself, its consequences in terms of damage and losses and the countermeasures to be undertaken to hamper such negative effects.
Unfortunately, though damage and ground motion scenarios have been largely developed in recent years (Fäh et al., 2000; Dolce et al., 2003), earthquake induced ground failure scenarios have been generally disregarded under the general scope of assessing seismic shaking and directly related losses.
Among the few attempts to afford the specific topic of assessing seismically-induced landslide scenarios we can mention the pioneer work by Wieczorek et al. (1985) and, more recently, the work Jibson et al. (2000).
In order to evaluate the most landslide-prone areas and the critical aspects of a scenario characterised by a quasi-simultaneous seismic triggering of numerous mass movements, some methods have been proposed in the scientific literature for regional scale analysis of earthquake-induced landslide hazard (Keefer & Wilson, 1989; Harp & Wilson, 1995). A recent approach devised by Jibson et al. (1998) proposed to evaluate the areas exposed to ground failure phenomena in a seismic event scenario on the basis of topographic, geological and geotechnical data integrated in a GIS together with estimates of the seismic shaking expected. This approach was tested also by other authors and further developments were attempted for national scale hazard maps (Romeo, 2000), or to define the strength (measured by critical acceleration) required for slopes of a region to keep seismic failure probability within a given limit in a fixed time interval (Del Gaudio et al., 2003; Del Gaudio & Wasowski, 2004a).
A general problem arising in the application of these techniques is the evaluation of the shaking expected along landslide-prone slopes for a seismic event of given characteristics. The attenuation relationships reported in literature have been calibrated on accelerometric data acquired on sites whose characteristics are far from those of unstable slopes, considering that permanent stations are almost never located in such unstable conditions and that tests with temporary stations conducted during short time intervals seldom record significant seismic events.
Experiences gained abroad and in Italy suggest that earthquake-induced landslides may involve both rock masses and sandy or clayey soils in different ways (Rodriguez et al. 1999; Prestininzi & Romeo, 2000).
The collection and analysis of seismically induced landslide data set at a global scale have already allowed the estimation of relationships between the occurrence of a landslide and some characteristics of the inducing earthquake, such as epicentral distance and magnitude (Rodriguez et al., 1999); pseudodynamic analysis performed at a regional scale also allowed the depiction of earthquake-triggered landslide scenarios.
The scientific literature offers a restricted number of papers on earthquake-induced landslides in natural clayey slopes. Experiences abroad refer mainly to the 1964 Alaskan earthquake (MS=8.5) which triggered the Turnagain Heights (Seed & Wilson, 1967) and the Fourth Avenue (Seed, 1968) landslides, whose failure mechanism, initially attributed to the liquefaction of the sandy layers, was subsequently interpreted in terms of progressive degradation of the undrained strength of the clays (Stark & Contreras, 1998). More recent landslides are those in two clayey slopes induced by the 1988 Saguenay earthquake in Canada (MS=5.9) (Lefebvre et al., 1992); considering the great distance of the landslides from the epicentre of the earthquake (some 175 km), the Authors point out that local amplification phenomena may have been critical factors in triggering the slides.
Many other case-histories are also available in both international and Italian literature; among them it is worth quoting the 1995 Kobe earthquake (Sassa et al., 1996), the 1989 Loma Prieta earthquake (Keefer, 1998), the 1980 Irpinia earthquake (Hutchinson & Del Prete, 1985; D'Elia et al., 1986; Martino & Scarascia Mugnozza, in press), the 1997 Umbria-Marche earthquake (Bozzano et al., 2001), the 2001 El Salvador earthquake (Evans & Bent, 2004), the 2002 Palermo earthquake (Bonci et al., 2004) and the 2002 Molise earthquake (Bozzano et al., 2004). Particularly remarkable is the epicentral distance of landslide event when compared with the earthquake magnitude and this can be ascribed to local site conditions (tectonic elements, stratigraphic conditions, morphological characteristics) which induce focalization of the seismic input.
The possibility of carrying out reliable a posteriori reconstructions of the mechanism of earthquake-induced landslides depends, among other things, on an adequate knowledge of the cyclic characteristics of the soil for a wide range of shear strains. Indeed, the magnitude of the involved shear strains often requires knowledge of the stiffness and damping characteristics in non-linear field. This is particularly true for scaly clays, for which there are few experimental data (e.g. D'Elia, 1983; Olivares, 1996; Olivares & Silvestri, 2001), which is absolutely insufficient for gaining a comprehensive picture of the stress-strain behaviour of these geological materials in the presence of an earthquake. Indications from the literature suggest that stiffness and damping, as well as the threshold shear strains (linear, and volumetric), are influenced by the meso-structural properties of these soils rather than by the plasticity of each clay fragment and that the non-linear behaviour is pronounced even at small-to-medium strain levels (Lanzo, 1993). The correct knowledge of the experimental laws governing the reduction of stiffness and increase in the damping ratio with shear strain may have significant effects on the field of acceleration that occurs within a slope, and hence on local amplification phenomena, if any, that are strictly dependent on the characteristics of the materials in which seismic waves propagate.
With reference to these findings, the study of local seismic response in landslide areas is currently a topic of great scientific interest, which can be widely applied to both landslide risk assessment and seismic microzonation.
A local amplification of seismic ground motion can indeed give rise to landslide coseismic trigger and, as a consequence, can increase the seismic hazard of a site.
Since the end of the '80, several methodologies aiming at the evaluation of local seismic response were developed. The analysis of the H/V components spectral ratios of the motion is a widely applied technique, due to its low cost and speed of use. This technique was proposed by Nakamura (1989) to assess the amplification of a site by microtremor records (ambient noise of natural or human source).
When velocimetric or accelerometric records of seismic events are available, the greater problem to face in order to assess the site response is to remove the effects due to source and wave path. Up to now, the best approach is that proposed by Borcherdt (1994): it takes into account the ratio between the Fourier spectra of the records obtained at the site and the spectra of the records of the same event obtained in a "reference station".
At present, the methodologies addressed to the analysis of local seismic response have been seldom applied to study both seismic trigger of landslides and site amplification in landslide areas. In the last years some experiments on this topic were performed by Havenith et al. (2003) in Kyrgyzstan, on rock-avalanche deposits. In Italy, the first results obtained by a research project financed by the European Union were presented (Del Gaudio & Wasowski, 2004b); the research aims at the accelerometric monitoring of an area which is characterised by widespread landslide phenomena and high seismic hazard. Also in Italy (Cavola locality, in Emilia Romagna region), the velocimetric monitoring of a landslide known for several earthquake-induced reactivations is being performed (Bordoni et al., 2005).
The results obtained for the Caramanico Terme landslide until now do not yet allow to draw conclusions of general validity for the following reasons: a) the local geological and geomorphological situation is rather complex and requires the collection of further data to distinguish the influence of different factors (lithological and topographical) on the observed amplification effects; b) acceleration time-histories from a station that can be reliably assimilated to a reference ground (with a sufficiently flat and rigid ground) have not yet been recorded; c) the azimuthal distribution of the recorded events is rather limited and this makes difficult to separate the site effects from the effects due to possible directivity of seismogenic sources or peculiarities of the source-site propagation path; d) the quality of data on the physical properties of soils is not sufficient for a reliable numerical modelling of the site response to seismic shaking.
The results of previous local seismic response studies lead to deem of particular importance for the seismo-induction of landslides the amplification effects of ground motion due to the impedance contrast between bedrock and landslide deposit (Borcherdt, 1994; Nakamura, 1989), to topographical effects (Chavez-Garcia et al., 1997) and to trapped wave modes in fault zone areas (Li et al., 1996; Rovelli et al., 2002).
Different geophysical techniques have been applied to landslides occurred in different geological formations and scenarios. In most cases geophysics has been used to infer about the geometry of the landslide body.
Recently, the approach which is proving to be successful is the integrated use of different geophysical survey techniques (Jongmans, 2000; Cardarelli et al. 2004).
Several geophysical techniques can also be successfully used in the determination of low strain dynamical properties of the geological units involved in the landslide (Rix, 1988).
These techniques, within the frame of the definition of the geotechnical behaviour of these materials, are capable to investigate large volumes of virtually undisturbed soil and thus allow to show the influence of the macro-structure on the stiffness parameters at low-strain, on the variation of these parameters within the soil, giving also further insight about the water circulation in the subsoil.
In conclusion, it must be stressed that any comprehensive assessment of seismic hazard and risk should include the hazard associated with earthquake-induced landslides. Within this frame, investigations on mechanisms leading to co- and post-seismic slope failures, with reference to specific site conditions focussing the seismic waves, should be furtherly developed.
This is the basic rationale of the present research proposal. <<<