RICERCA ITALIANA     

Analysis and mitigation of the risk induced by fast slope movements.

Università degli Studi di Parma

Abstract

The detaching of rock blocks from the mountainsides is a phenomenon commonly spread in every mountanious region of the world as well as along the rocky coastlines. The evolution of such landslide phenomena causes rock falls, rock avalanches, debris flow and other type of fast movement that are constituting a constant hazard for lives, structures and communication lines. The defense of such valuable items can be accomplished through the stabilization of the rock cliffs (using active interventions) or through the protection of human lifes and structures by means of passive interventions. The design of both type of interventions requires the knowledge of the geometrical, mechanical and hydraulical features of the rock masses forming the cliffs or the mountainsides. The first objective of this research program is to improve the methodologies and techniques for the acquisition of the rock mass discontinuities geometrical, mechanical and hydraulical features which, is known, have a major role in generating such hazardous phenomena. The aim of the research carried out by the Parma Unit, is to develop and test a procedure that will automatically select from the point cloud determined on the rock face a set of points distributed on a particular discontinuity, location, dip and dip direction using the least squares estimate of the plane interpolating the set of points. Likewise, the normal vector to the surface may be computed from an interpolation or approximation of the surface by appropriate functions. To become a real alternative to a traditional survey, interactive or automated software tools are necessary, to allow the efficient selection of the point sets on the discontinuities or the interpretation of the normal vector pattern. Few application of these interactive tool, still under development, were presented in recent papers by some of the people composing the Parma research unit. A further progress will be related to the automation of the stability analysis phase that, starting from the survey output and from the automatically obtained geostructural model, will produce an hazard map of the surveyed area and will provide unstable volume calculation and position of the most likely detachment zones. This automated system will provide all the parameters needed for the propagation analysis of the rock fall and debris flow phenomena. The University of Milan research Unit, will focus its work on the analysis of the structure of spatial correlation that are mainly affecting the hydrogeological behaviour of the discontinuities and to evaluate the effect of groundwater influence on rock masses stability. In the first year the first step of the research will be to organise a database of available technical information in the pilot area in Valchiavenna to evaluate their spatial distribution. All the information will be managed within a geographical information system. The survey of springs in the area will also be performed also considering hydrochemical and discharge data already available. The preliminary spatial analysis of site specific data and the overlay with the geological and structural features of the area will consent to identify the most susceptible sectors in terms of stability, considered as pilot areas, where the following technical activities will be focused. This part of the research will be done in collaboration with Unit 1 of the Università di Parma using the remote sensing techniques to survey discontinuities in rock masses and the automatic elaboration of the main failure kinematics system developed by Unit 1. In the selected pilot areas other detailed structural surveys will be performed to complete eventual lacks of technical information for each lithological and structural unit. The first objective of the study will be to highlight how to integrate remote and on site techniques to obtain rapid and reliable estimate of area most susceptible to rockfalls. The second aspect of this research program is tied to the stability analysis of the potentially unstable areas located through the on site surveys. These analysis are constituting the primary objective of the Research Unit of Torino 1 that will analyze, using already tested numerical programs for the verification of potential sliding and falling kinematisms of rock blocks. The Turin 1 unit will carry out the analysis of the run out path of a single or several blocks in order to define the most probable areas that could be involved in the deposition of falling rocks or rock avalanches. These analysis have the objective of facilitate the decisions related to the type and location of the most effective intervention for the mitigation of these landslide phenomena. The Unit Turin 2, has the objective of determining the invasion areas of debris flow phenomena. The Unit will carry out several back analysis of already known debris flow in order to calibrate, using two different numericla model, the specific features of these type of phenomena. <<<

Principal Investigator

Gianpaolo Giani Università degli Studi di PARMA

Research Objectives

The final objective of this Research Project can be defined considering the results and the objectives accomplished during the previous PRIN 2005 titled “Theoretical and experimental studies for the triggering and evolution of rockfall phenomenom” in which some of the Units partecipating to this project were also involved.
Starting from these premises it can be said that the most common fast movements (falls, rock avalanches , debris avalanches and debris flows) have been treated and considered.
The global objective of this project is therefore to better the knowledge of the phenomena that are determining the triggering and the various types of evolution of the failure of portions of rock mass detached from rock cliff or mountainside.
In particular the innovative aspects, that are also constituting the objectives of this research project are summarized in the following few pharagraphs.
Estimate of the reliability and advantages of the remote sensing survey techniques (photogrammetry and/or laser scanner) for the definition of rock mass geometry. Geometrical reconstruction of rock masses and estimate of the block volumes and of the grain size distribution of the debris deposits (Parma Unit).
Development of a methodology for the evaluation of the recurrence probability of rock fall phenomena; definition of the rock fall paths and evaluation of the rock mass motion features, in order to define the most probable invasion areas, the appropriate parameters for the design of defensive systems and the production of hazard maps for the valleys located at the foot of the slopes subjected to rock fall (Turin 1 Unit).
Determination of the rheologic behaviour of the natural materials (debris and, in suborder, sands and fine soils) constituting the body of a debris flow. Definition of the parameters to be used for the numerical models of debris flows. Calibration of these parameters using a back analysis tehnique, comparing the results of the numerical modeling with the experimental evidences gained from on site surveys and from laboratory testing (Turin 2 Unit).
Development and calibration of a geostatistical model able to evaluate the spatial distribution of geometrical and physical features of the rock mass discontinuities. Estimate of the hydraulic conductivity of the rock mass and of its primary discontinuities with the support of simple on site hydraulic measurements and of the geostructural surveys and statistical analysis results. Development of a 3D flow model within a fractured rock mass (Milan Unit). <<<

First Results

The results of this Research Project can be defined considering the results and the objectives accomplished during the previous PRIN 2005 titled “Theoretical and experimental studies for the triggering and evolution of rockfall phenomenom” in which some of the Units partecipating to this project were also involved.
Starting from these premises it can be said that the most common fast movements (falls, rock avalanches , debris avalanches and debris flows) have been treated and considered.
The global objective of this project is therefore to better the knowledge of the phenomena that are determining the triggering and the various types of evolution of the failure of portions of rock mass detached from rock cliff or mountainside.
In particular the innovative aspects, that are also constituting the objectives of this research project are summarized in the following few pharagraphs.
Estimate of the reliability and advantages of the remote sensing survey techniques (photogrammetry and/or laser scanner) for the definition of rock mass geometry. Geometrical reconstruction of rock masses and estimate of the block volumes and of the grain size distribution of the debris deposits (Parma Unit).
Development of a methodology for the evaluation of the recurrence probability of rock fall phenomena; definition of the rock fall paths and evaluation of the rock mass motion features, in order to define the most probable invasion areas, the appropriate parameters for the design of defensive systems and the production of hazard maps for the valleys located at the foot of the slopes subjected to rock fall (Turin 1 Unit).
Determination of the rheologic behaviour of the natural materials (debris and, in suborder, sands and fine soils) constituting the body of a debris flow. Definition of the parameters to be used for the numerical models of debris flows. Calibration of these parameters using a back analysis tehnique, comparing the results of the numerical modeling with the experimental evidences gained from on site surveys and from laboratory testing (Turin 2 Unit).
Development and calibration of a geostatistical model able to evaluate the spatial distribution of geometrical and physical features of the rock mass discontinuities. Estimate of the hydraulic conductivity of the rock mass and of its primary discontinuities with the support of simple on site hydraulic measurements and of the geostructural surveys and statistical analysis results. Development of a 3D flow model within a fractured rock mass (Milan Unit).
The interest for the advancement of the knowledge and the application potentials of the results of this project are hereby summerized.
The use of laser scanner and/or photogrammetry techniques is growing in popularity and is becoming more and more widespread in the rock mechanics field. At the 2007 International Congress of Rock Mechanics held in Lisbon last July, a special workshop focusing on the theme of this research has been organized and supervised by Prof. H. Einstein. During these sessions several presentation were dealing with the application of these remote sensing techniques to the rock mechanics field. Even during the last US symposium on Rock Mechanics several papers dealing with this subject were presented.
The geotechnical group of the University of Parma has been, in these last few years, very active in the study and application of these methodologies and has all the intentions in continuing the work done so far.
The interest for the application of these new methodologies is grown since few years ago, particularly for the advantages that these techniques present when compared with traditional geostructural surveys: higher precision, lower subjectivity, possibility of application in areas not easily accessible and ability of constructing large data bases that can be processed in more confortable locations (i.e. office).
The rock mass tessellation methodologies can be applied with higher reliability having complete sets of orientations, persistance and spacing data from the discontinuities that were surveyed withphotogrammetry and/or laser scanners.
The construction of a grain size distribution curve of an accumulated debris carried out using these remote sensing techniques constitutes an extremely innovative aspect of this research.
The development of a methodology for the evaluation of the temporal occurence of rock fall phenomena and for the definition of the rock block travelling paths, as well as, of the other rock fall motion features will allow for the correct determination of areas of invasion, rock fall protection system design parameters and definition of hazard maps and, therefore, has an important applicative fall back.
The determination of the temporal occurrence of a rock fall phenomena is a very ambitious objective and this research results could produce only the first indicative answer to the problem. On the other hand, the results obtained by the kinematic analysis, stability analysis and rock fall analysis for the determination of the invasion areas will constitute an innovative aspect in the field of the rock fall hazard and risk studies carried out at a basin scale.
The studies on the evolution mechanisms of debris flows are carried out, since many years ago, by several researchers in the fields of physics, hydraulic enginnering and geotechnical engineering. The provvisional analysis of the debris flow path and the estimation of invasion areas are frequently performend by the Civil Protection Governmental Offices using 2D models, with very simple motion parameters derived from symplified assumptions. Some of the goals of this research program are to better the knowledge of the various rheological behaviour of different debris masses when subjected to flow, to better the knowledge of the mechanical features that are governing the phenomena and, at last, to develop a 3D model for the analysis of the depositional phase of the phenomena. The prectical fall back of these studies it is of extreme importance for the study and the design of territory protection systems.
Many are the studies, recovered from the scientific literature, regarding the use of the statistic science for the evaluation of the geometrical features of the discontinuities. The fractal method and the regionalized variable theory have minor response and are mostly used for the analysis of the discontinuities surface roughness. The use of the geostatistical approach for the distribution analysis of geometrical features, such as the aperture and spacing of discontinuities, is an innovative aspect of this research program as well. The results of these part of the research will have primarily a scientific value.
Again, the determination of the hydraulic conductivity of a single discontinuity or of a discontinuos rock mass is one of the most difficult tasks in rock mechanics, expecially when starting from on site surveys and simple hydraulic measurements (not considering the pumping tests). The objective of the research in this field is a tentative, the results of which, if validated by on site surveys, could direct such studies toward more simple and economic survey and analysis methodologies, with important practical fall backs.
Finally, the evaluation of hydraulic pressures at the onset of the potential sliding surfaces is one of the most important aspects concerning the stability analysis of rock slopes. The development of a 3D flow model of a rock mass constitutes an objective that could give very interesting scientific fall backs, as well. <<<

Timescale

24 months

National and international background

Rock fall phenomena are common throughout all mountainous regions and along coastal cliffs, constituting a constant danger for lives, property, and human activities. Of the approximately 34,000 landslide phenomena recorded in Piemonte by the IFFI project (ARPA Piemonte, 2004), 20.9% were made up of single rock falls or areas susceptible to diffuse falls. In spite of such a wide diffusion of the phenomenon, the literature contains few examples of analyses carried out on a basin scale. In fact, the process, even though apparently traceable to a rather simple scheme of the mechanisms that govern it, is complicated by the high degree of uncertainty in the definition of the parameters that control both the triggering and the run-out phases. Furthermore, the wide extension of the areas involved (consider the maps of major transportation systems like motorways and railways) makes it difficult to collect sufficient information to conduct detailed analyses. One of the first examples of methodologies dedicated to the definition of rockfall hazard in vast areas is the STONE computer code (Guzzetti et al., 2002; Crosta &amp; Agliardi, 2003), which is capable of evaluating the invasion area and producing thematic maps on the basis of the topographical and geomorphological information available on a basin scale and using a 3D simulation model of the trajectories of falling masses. Other similar methods have been proposed by many authors (e.g. Wieczorek et al., 1998; Jaboyedoff &amp; Labiouse, 2003). A limitation of these approaches is due to the absence of a mechanical model that allows the evaluation of the possibility of a detachment of rock volumes from the cliffs; in fact the codes are not capable of taking into account the structural conditions of the slope and limit the probabilistic evaluation to the evolution phase of the phenomenon, for which it is sufficient to know the morphology of the slope, its lithological conditions, and the land use (essential for the evaluation of the impact damping coefficient). In rigorous general terms, the mitigation of the risk when dealing with landslide phenomena should be linked to the calculation of the hazard, as a function of the extent of the involved area, the intensity of the phenomenon, and the probability of occurrence (Bonnard et al., 2004). Moreover, the quantitative forecasting of rock falls is complicated because the detachment may be caused by a combination of minor variations in the geometrical, mechanical and hydraulical parameters of the discontinuities that play a crucial role for the formation of the blocky rock mass and for its stability. These parameters are usually obtained from detailed on site geostructural surveys performed on outcropping rock masses. In the last 30 years new techniques of immage acquisition have been developed and, when appropriately targeted, these remote sensing techniques could constitute a valid alternative to the classical geostructural surveys. As first approximation, these techniques can be divided in two groups: monoscope and stereoscope. Images recorded on site are a primary source of information: photo analysis or photogrammetric methods, complementary or alternative to the compass survey, have been developed in the last three decades. Broadly speaking, they can be divided in monoscopic or stereoscopic. In both approaches, partial or full automation of feature identification and measurements is pursued. The relationship between discontinuity sets in terms of joint hierarchy and different kinds of joint terminations can be observed on images as discontinuity traces. Tsoutrelis et al. (1990) and Crosta (1997) build manually discontinuity trace maps from single images. A real step ahead was performed with the development of method for automatic recognition of discontinuities on images. Reid et al. (2000) identify traces as ravines in the image brightness surface, classifying pixels based on gradient and curvature analysis and using a line following algorithm from seed points manually selected. Hadjigeorgiou et al. (2003) review previous work on automatic trace detection and compare the performance of several edge and line detection algorithms. Overall, poor reliability of the methods was highlighted as well as the great dependence of results on rock texture, illumination conditions and threshold settings, resulting in meaningless segments or in excessive fragmentation. To solve the ambiguities, Lemy and Hadjigeorgiou (2003) measure several geometric and radiometric parameters of the image segments, assuming discrimination in parameter space to be achievable. To this aim, they use neural networks, reporting a good success rate; however, it is not clear whether training need to be repeated in every project (or even in any image). Kemeny and Post (2003) derive 3D fracture orientations from 2D fracture trace information gathered from digital images. It is assumed that the orientations of the rock face and the camera are known, that the rock is relatively flat, and that the camera is positioned perpendicular to the strike of the rock face. The method estimates the constant K and the mean value of a Fisher probability density function for joint orientation data; from synthetic data it has been shown that the accuracy of dip and dip direction is very good (about 1°). Traces are delineated manually or semi-automatically by extracting lines with the well-known Hough transform. Stereoscopic methods combine information from two pictures of the same rock face, allowing direct extraction of 3D information (Hagan, 1980, Harrison, 1993). Gaich et al. (2004 and 2006) use stereo photogrammetry with a digital panoramic camera to maintain high resolution on the object and less expensive than off-the-shelf digital frame cameras for small areas. After the DTM (Digital Terrain Model) of the rock surface is generated using triangulation, images are draped on the DTM and a graphical interface allows the user to measure position, dip and dip direction of discontinuities as well as distances and areas. The surface normal in a given position is computed from the mean orientation values of the neighbouring surface elements. Discontinuity orientation from traces can be measured if the corresponding 3D polyline effectively defines a plane. Normal stereo imaging geometry is used, that does not yield homogeneous accuracy and might have unfavourable angles with respect to surface normal, unless simple rock shapes are imaged. Measurements are interactive only, which is not very efficient when complex rock structures have to be sampled. Birch (2006) discusses the application of photogrammetry to rock face characterization, stressing the need for appropriate image block design and accuracy issues. Automatic image orientation and generation of DTM mass points by digital image correlation is reported, but automation of discontinuity extraction is not mentioned. In the last years, laser scanning (Kemeny et al., 2003; Dolan, 2003; Feng, 2003) has emerged as an alternative to photogrammetry to acquire dense point clouds on rock faces. The methodology applied in this project, already presented by some participant of the Parma research unit and still under development, is based on a RANSAC algorithm. RANSAC is used for image analysis and automated cartography to search a predefined model in a particular sets of experimental data. In the case of a portion of DSM corresponding to a group of planes, the applied algorithm allows for the localization of the several planes and the estimate of their equations, even when these are affected by a relevant percentage of errors. The evaluation of strengthness and deformability and hydraulic properties of rock masses and their spatial distribution play a very important role in the geomechanical characterization of slopes to perform stability analysis and to eventually design the appropriate stabilization techniques (Hudson e al. 1997, Hoek 2007). The in situ collection of data and measures of parameters influencing the qualitative characteristics of rock masses is often limited by the lack of rock outcrop and by accessibility to slopes. In these cases data availability does not allow the estimation of rock masses technical parameters for the whole study area (Baecher 1983). If some of the data can be nowadays partially surveyed through the use of remote sensing techniques, data of discontinuities characteristics (opening, roughness, filling, humidity conditions) always need their direct measures in situ through more or less conventional techniques (Priest 1993, ISRM 1978). In any case it is possible to analyse the statistical properties of the main parameters of discontinuities together with their structure of spatial correlation through experimental variograms (Samper e al. 1989) by interpreting them as random function in the space. Thematic maps for each parameters can then be produced through a kriging analysis also evaluating the degree of uncertainty of the results by the variance of the errors maps (Guadagnini e al 1998). This approach is very useful for all parameters affecting water movement in rock masses which represents a main topic in their stability analysis for two main reasons:a) water presence in discontinuities has a great influence on their shear strength; b) hydraulic conductivity of rock masses is often an expression of their structural looseness (Trimmer, e al.. 1980). That's why a correct evaluation of hydraulic conductivity through robust and reliable approaches represents a very important step to be developed for studies of rock slope stability. Statistical analysis of data directly surveyed on site allows to relate parameters as opening, roughness, persistency, spacing of discontinuities, with a critical state representing the minimum structural condition allowing water movement in the rock mass. From here this parameters can be use to estimate hydrogeological characteristics of rock masses such as hydraulic conductivity and porosity (Berkowitz, 1995, Charlaix e al. 1987, de Dreuzy e al., 2001 a,b, Ge, S. 1997, Oron e al. 1998). These data can then be verified and validated through a direct approach mainly based on permeability tests, giving results with a very local validity, and a indirect and more qualitative but less influenced from transient site and boundary conditions approach, based on hydrochemical and hydrological characteristics of springs, main superficial expression of groundwater movement in rock masses in the area. The determination of the most appropriate spatial correlation for each factor allow the definition within rock masses of sector, named hydrogeological unit, showing similar hydrogeological behaviour which do not necessary coincide with the lithological units. The hydrogeological unit are the base of the three-dimensional conceptual hydrogeological model which must necessary be developed and build for the implementation of a reliable numerical model for groundwater analysis in rock masses. The model must then be calibrated for well known hydrologic conditions (Cacas, e al. 1990) and successively used to analyse different critical scenarios of aquifer recharge to determine the most hazardous ones for the development of rock mass instabilities. The study will be developed in a pilot area in Valchiavenna (SO), where the group of Engineering Geology of the Università degli Studi di Milano has a local department since 2000 for the study of the Alpine environment. The analysis of the evolution of rock fall phenomena, rock avalanches and debris flows, together with other fast movements, plays an important role in order to reduce the hazard induced by those phenomena. The main problem resides in the determination of the invasion areas that could be occupied by the fallen material and in the assesment of the motion path and trajectories that could, in turn, give the correct parameters for the design of protection and defense systems. Lately, the rock fall and debris flow phenomena are commonly studied through the application of numerical and analytical modelling. Often, with the help of on site tests, it is possible to calibrate the motion parameter by means of a comparison between experimental data and calculation results obtained from known phenomena. Debris flows are natural phenomena characterised by great volumes of concentrated mixtures of water and sediments. They develop in mountainous areas with high velocities and are triggered by unusual presence of water. A further classification is based on the grain size: muddy debris flows that are characterised by a well sorted grain distribution and high fine content; granular debris flow that are poor in fine particles. Sediment concentration and solid material properties are among the key elements influencing the rheological or flow resistance characteristics of debris flows (Pierson and Costa, 1987), which may vary during a given event. Rheological behaviour of concentrated suspensions has been faced by three, apparently distinct, fields, which consider different types of material under various conditions: rheology of suspensions; physics of granular matter; geotechnics. Although a few methods of analysis are based on relatively rigorous approaches (Pastor et al. 2006), nevertheless the study of debris flows is generally performed by simplified models due to the complexity of the natural process. Campbell (2006) by analysing granular flows by a general point of view observes: "Yet the design of granular systems is still something of a black art, in part because even the most basic flow mechanisms of granular material are not well understood. In fact science has not identified the set of material properties that control the flow behaviour". In spite of these uncertainties concerning ideal materials (spheres, monodisperse suspensions, etc.) under ideal conditions, land protection needs tools in terms of risk mitigation. This consideration leads to the use of very simples models based on few parameters, allowing the assessment of run out and velocity of debris flows (Deangeli, 2007). The models applied for the study of real debris flows use techniques developed for analysis of the flow of fluids in open channels (Hungr, 2002). The main problem is the choice of an equivalent fluid whose rheological properties supply a bulk behavior of the flow similar to the bulk behavior of the actual granular flow. Consequently the properties of the equivalent fluid do not necessary correspond to those of the actual flowing material. The system is generally considered single phase and the adopted rheological laws are: Newton, Bingham, Bagnold e Voellmy. In the Italian context the assessment of debris flow risk is generally performed by commercial code (DAN, FLO-2D), based on the aforementioned constitutive behaviour. Several studies compares the results obtained by back analysing debris flow by the use of different models (Tecca et al. , 2006; Sosio et al., 2006). The critical aspect of these modelling is the calibration of rheological parameters of the equivalent fluid, required to fit in site data (Rickenmann et al. 2006). Geological background supplies fundamental indications on the physical characteristics of sediment water mixtures e on the probable velocity distribution in different geomorphologic contexts. A general criterion to distinguish the predominant behaviour of suspensions is generally performed by the use of dimensionless numbers (Reynolds, number, Bagnold number, etc.), which are based on physical features of the solid phase and the velocity of the bulk debris flow. <<<