Ph.D. Dissertation on Landslide Hazard and Risk Assessment

The Title pages ...

The Table of Content ...

Landslides play an important role in the evolution of landforms and represent a serious hazard in many areas of the World. In places, fatalities and economic damage caused by landslides are larger than those caused by other natural hazards, including earthquakes, volcanic eruptions and floods. Due to the extraordinary breadth of the spectrum of landslide phenomena, no single method exists to identify and map landslides, to ascertain landslide hazards, and to evaluate the associated risk. This work contributes to reduce this shortcoming by providing the scientific rationale, a common language, and a set of validated tools for the preparation and the optimal use of landslide maps, landside prediction models, and landslide forecasts. I begin the work by critically analysing landslide inventories, including archive, geomorphological, event and multi-temporal maps. I then present methods to analyse the information shown in the inventories, including the assessment of landslide density and spatial persistence, the completeness of the landslide maps, and the estimation of the recurrence of landslide events, the latter based on historical information obtained from archive or multi-temporal inventories. I then use statistical methods to obtain the frequency-size statistics of landslides, important information for hazard and risk studies. Next, I discuss landslide susceptibility zoning and hazard assessment. I examine statistical and physically-based methods to ascertain landslide susceptibility, and I introduce a scheme for evaluating and ranking the quality of susceptibility assessments. I then introduce a probabilistic model to determine landslide hazard, and I test the model at different spatial scales. Next I show how to determine landslide risk at different scales using a variety of approaches, including probabilistic methods and heuristic geomorphological investigations. Risk evaluation is the ultimate goal of landslide studies aimed at reducing the negative effects of landslide hazards. Lastly, I compare the information content of different landslide cartographic products, including maps, models and forecasts, and I introduce the idea of a landslide protocol, a set of regulations established to link terrain domains shown on the different landslide maps to proper land use rules. I conclude the work by proposing recommendations for the production and optimal use of various landslide cartographic products. The recommendations and most of the results shown in this work are the results of landslide hazard research conducted in central and northern Italy. However, the lessons learned in these areas are general and applicable to other areas in Italy and elsewhere.

A “landslide” is the movement of a mass of rock, debris, or earth down a slope, under the influence of gravity (Nemčok et al., 1972; Varnes, 1978; Hutchinson, 1988; WP/WLI, 1990; Cruden, 1991; Cruden and Varnes, 1996). Different phenomena cause landslides, including intense or prolonged rainfall, earthquakes, rapid snow melting, and a variety of human activities. Landslides can involve flowing, sliding, toppling or falling movements, and many landslides exhibit a combination of two or more types of movements (Varnes, 1978; Crozier, 1986; Hutchinson, 1988; Cruden and Varnes, 1996; Dikau et al., 1996). ... The range of landslide phenomena is extremely large, making mass movements one of the most diversified and complex natural hazard (Figure 1.1). Landslides have been recognized in all continents, in the seas and in the oceans. On Earth, the area of a landslide spans nine orders of magnitude, from a small soil slide involving a few square meters to large submarine landslides covering several hundreds of square kilometres of land and sea floor. The volume of mass movements spans sixteen orders of magnitude, from a single cobble falling from a rock cliff to gigantic submarine slides. Landslide velocity extends at least over fourteen orders of magnitude, from creeping failures moving at millimetres per year (or even less) to rock avalanches travelling at hundreds of kilometres per hour. Mass movements can occur singularly or in groups of up to several thousands. Multiple landslides occur almost simultaneously when slopes are shaken by an earthquake or over a period of hours or days when failures are triggered by intense or prolonged rainfall. Rapid snow melting can trigger slope failures several days after the onset of the triggering meteorological event. An individual landslide-triggering event (e.g., intense or prolonged rainfall, earthquake, snow melting) can involve a single slope or a group of slopes extending for a few hectares, or can affect thousands of square kilometres spanning major physiographic and climatic regions. Total landslide area produced by an individual triggering event ranges from a few tens of square meters to hundreds of square kilometres. The lifetime of a single mass movement ranges from a few seconds in the case of individual rock falls, to several hundreds and possibly thousands of years in the case of large dormant landslides.

In this chapter, I describe the study areas where the research illustrated and discussed in the following chapters was conducted. For each area, I provide general information on the type and abundance of landslides and on the local setting, including morphology, lithology, structure, climate, and other physiographic characteristics. For some of the areas, I give information on the type and extent of damage caused by the slope failures. Where appropriate, I provide a brief description of the topographic, environmental and thematic data used to perform landslide susceptibility zonings, landslide hazard assessments, and landslide risk evaluations. ...

Any serious attempt at ascertaining landslide hazard or at evaluating landslide risk must begin with the collection of information on where landslides are located. This is the goal of landslide mapping. The simplest form of landslide mapping is a landslide inventory, which records the location and, where known, the date of occurrence and types of landslides that have left discernable traces in an area (Hansen, 1984; McCalpin, 1984; Wieczorek, 1984). Inventory maps can be prepared by different techniques, depending on their scope, the extent of the study area, the scales of base maps and aerial photographs, the quality and detail of the accessible information, and the resources available to carry out the work (Guzzetti et al., 2000). In this chapter, I first critically discuss the various types of landslide inventories and the methods and techniques used to prepare them. Then, I present landslide inventories of different types and scales prepared for Italy, the Umbria Region, and for selected areas in the Umbria Region, including the Collazzone area. ...

The information shown on landslide inventories can be used for a variety of analyses, including: (i) investigating landslide spatial abundance, through the production of landslide density maps; (ii) comparing inventory maps obtained from different sources (e.g., archive and geomorphological) for the same area; (iii) evaluating the completeness of the inventories; (iv) ascertaining landslide geographical persistence, by comparing event and geomorphological inventories; (v) estimating the frequency of slope failure occurrence, by analysing historical catalogues of landslide events or multi-temporal inventory maps; (vi) obtaining the statistics of landslide size; (vii) ascertaining landslide susceptibility and hazards, including the validation of the obtained susceptibility and hazard forecasts; (viii) determining the possible impact of landslides on built-up areas or the infrastructure; and (ix) contributing to establish levels of landslide risk. The quality and reliability of the different analyses obtained from a landslide inventory depend largely (often entirely) on the quality and completeness of the original landslide map. For this reason, one should always: (i) aim at compiling accurate and precise inventories, (ii) document the sources of information used to obtain the inventories, (iii) accurately describe the techniques, methods and tools used to prepare or compile the inventories, and (iv) try to assess the completeness of the obtained inventories. Limitations of landslide inventories should always be known (i.e., explicit and clear) to the users of the maps or the archives. In this chapter, I discuss some of the possible applications of landslide inventories. I first demonstrate the construction and use of landslide density maps. I then show methods to compare geomorphological and historical inventories. I discuss an index to quantify the degree of matching between inventories, and I show an application for the comparison of the three landslide maps available for the Collazzone study area. I further discuss the issue of the completeness of the landslide inventories, and I use two event inventories available for Umbria to investigate geographical landslide persistence. Finally, I show how to ascertain the temporal frequency of slope failures from archive inventories. ...

The size (e.g. length, area, volume) of individual landslides varies largely. As shown in Figure 1.1, the length of landslides varies from less than a meter to several hundreds or even thousands of kilometres for submarine slides. Landslide area spans the range from less than a few square meters, for shallow soil slides, to thousands of square kilometres, for large submarine failures. The volume of single mass movements ranges from less than a cubic decimetre, for rock fragments falling off a cliff, to several hundreds of cubic kilometres, for gigantic submarine slides (Locat and Mienert, 2003), or for slope failures identified on the Moon (Hsu, 1975), Mars (McEwen, 1989) and Venus (Malin, 1992). The frequency-size distribution of landslides is important information to determine landslide hazards (Guzzetti et al., 2005a) (see § 7.3), and to estimate the contribution of landslides to erosion and sediment yield (e.g., Hovius et al., 1997, 2000; Martin et al., 2002; Guthrie and Evans, 2004b; Lavé and Burbank, 2004). For these reasons, it is essential that the distributions are quantified precisely, using accurate and reliable methods. In this chapter, after a review of the limited literature, I show how to obtain frequency-area and frequency-volume statistics of landslides from empirical data obtained from landslide inventories. I then discuss applications of the obtained frequency statistics of landslides, with examples from the Umbria region, including an application to investigate the completeness of the three landslide inventory maps available for the Collazzone area. ...

In the literature, confusion exists between the terms landslide “susceptibility” and landslide “hazard”. Often, the terms are used as synonymous despite the two words expressing different concepts. Landslide susceptibility is the likelihood of a landslide occurring in an area on the basis of local terrain conditions (Brabb, 1984). It is the degree to which a terrain can be affected by future slope movements, i.e., an estimate of “where” landslides are likely to occur. Susceptibility does not consider the temporal probability of failure (i.e., when or how frequently landslides occur), nor the magnitude of the expected landslide (i.e., how large or destructive the failure will be) (Committee on the Review of the National Landslide Hazards Mitigation Strategy, 2004). In mathematical language, landslide susceptibility is the probability of spatial occurrence of slope failures, given a set of geo-environmental conditions. This is called “landslide analysis” by Vandine et al. (2004). Landslide hazard is the probability that a landslide of a given magnitude will occur in a given period and in a given area. Besides predicting “where” a slope failure will occur, landslide hazard forecasts “when” or “how frequently” it will occur, and “how large” it will be (Guzzetti et al., 2005a). Landslide hazard is more difficult to obtain than landslide susceptibility, as susceptibility is a component (the spatial component) of the hazard. More generally, landslide susceptibility consists in the assessment of what has happened in the past, and landslide hazard evaluation consists in the prediction of what will happen in the future. In this Chapter, I discuss landslide susceptibility zoning, whereas landslide hazard modelling will be dealt with in § 7. Here, I review the methods proposed to ascertain landslide susceptibility, including an analysis of the types of mapping units most commonly adopted, and of the relationships between the selected mapping units and the adopted susceptibility methods. I then examine a probabilistic model for landslide susceptibility, including problems and difficulties in its application, and I present an example of a landslide susceptibility model for the Upper Tiber River basin, an area that extends for about 4100 km2 in central Italy. Lastly, I discuss the problem of the verification of the performances of a landslide susceptibility model, including examples for the Collazzone area, in central Umbria. ...

A hazard is the likelihood that a danger will materialize. A natural hazard is the hazard posed by a potentially damaging natural event or process, such as an earthquake, flood, volcanic eruption, snow avalanche, hurricane, ground subsidence or mass movement. Landslide hazard refers to the potential for the occurrence of a damaging slope failure within a given area and in a given period. To properly define landslide hazard, the magnitude, size, or dimension of the expected failure must also be quantified, deterministically or in probabilistic terms, because the “magnitude” of the event is linked to its destructive power. Landslide hazard is portrayed on maps. A landslide hazard map partitions a territory based upon different levels of landslide hazard (landslide hazard zoning). As it will become clear later, producing a single landslide hazard map is problematic, as different hazard conditions (or probabilities) must be shown on the same map. An ensemble of maps can be prepared to show landslide hazard, and displayed in a GIS. In this chapter, I first examine a definition of landslide hazard, I then introduce a probabilistic model for landslide hazard assessment that fulfils the adopted definition, and I discuss known problems with the given definition and limitations of the proposed probability model. Next, I show three examples of the application of the proposed probability model for different types of landslides and at different scales, from the catchment to the national scale. In the first example, I illustrate an attempt to determine landslide hazard in the Staffora River basin (§ 2.6), exploiting a detailed multi-temporal inventory map and thematic information on geo-environmental factors associated with landslides. In the second example, I describe an attempt to determine levels of landslide hazard in Italy, based on synoptic information on geology, soil types and morphology, and an archive inventory of historical landslide events. In the third example, I exploit a physically-based computer model capable to simulating rock falls for the determination of rock fall hazard in south-eastern Umbria (§ 2.5). ...

Risk assessment is the final goal of many landslide investigations. It lays at the fuzzy boundary between science, technology, economy and politics, including planning and policy making. Assessing landslide risk is a complex and uncertain operation that requires the combination of different techniques, methods and tools, and the interplay of various expertises pertaining – among the others – to geology and geomorphology, engineering and environmental sciences, meteorology, climatology, mathematics, information technology, economics, social sciences and history. Despite the indisputable importance of landslide risk evaluation for decision making, comparatively little efforts have been made to establish and systematically test methods for landslide risk assessment, and to determine their advantages and limitations. In this chapter, after a brief review of the relevant literature, I present concepts and definitions useful for landslide risk assessment, including a discussion of the differences between quantitative (probabilistic) and qualitative (heuristic) approaches. I then make various examples of probabilistic, heuristic, and geomorphological landslide risk assessments. The examples include: (i) the determination of societal and individual levels of landslide risk in Italy, and a comparison with risk levels posed by other natural and man-made hazards, and by the principal medical causes of deaths in Italy, (ii) a preliminary attempt to establish the geographical distribution of landslide risk to the population in Italy, (iii) the determination of rock fall risk to vehicles and pedestrians along roads in the Nera River and the Corno River valleys, in eastern Umbria, (iv) the design and application of a geomorphological method for the determination of heuristic levels of landslide risk at selected sites in Umbria, based on information obtained from the interpretation of multiple sets of aerial photographs of different ages, combined with the analysis of historical information on past landslide events, and pre-existing knowledge on landslide type and abundance, (v) an attempt to determine the type and extent of landslide damage in Umbria, based on the analysis of a catalogue of landslides and their consequences, and (vi) an effort to establish the location and extent of sites of possible landslide impact on the population, the agriculture, the built-up environment, and the transportation network in Umbria. ...

The value of a map refers to its information content, which depends on the type of data shown, their quality and the extent to which the information is new and essential. A map is valuable when the data shown are useful to the user, i.e., when the map is both relevant and understood by the user (Guzzetti et al., 2000). A carefully designed inventory map that shows landslides as recognised by the interpreter, without any modification apart from scale or graphical constrains, is a basic map. A landslide density map obtained by interpolating an inventory map without any additional information is a derivative map. Landslide susceptibility and hazard maps obtained from an inventory are also derivative maps but, since they include additional information on factors such as lithology and morphology that are used to build the susceptibility or hazard models, they have an information content which is superior to that of the input maps, including the inventory. Risk assessments are complex, high level products that exploit basic, derivative and other thematic information and maps (Guzzetti et al., 2000). In this chapter, I first describe and compare the information content of different landslide cartographic products, including inventory, density, susceptibility and hazard maps, and risk evaluations. Next, I introduce and discuss the concept of a “landslide protocol”, i.e., a set of regulations established to link terrain domains shown on the different landslide maps to proper land use rules. ...

In this last chapter, I draw the conclusions and I propose general recommendations for the preparation and use of landslide inventory maps, of landslide susceptibility and hazard assessments, and of landslide risk evaluations. I draw my conclusions on what I have presented and discussed in the previous chapters, and I propose the recommendations based on the experience gained in landslide studies carried out mostly in Central Italy in the last twenty years. ...

I decided to enrol in a Ph.D. programme at the University of Bonn, in Germany, when I was well into my working carrier as a research geomorphologist in Italy. Embarking in such a venture required support from many, and particularly from my family and my closest colleagues. During this endeavour, I received constant encouragement and support from Emanuela and Martina. Their unconditioned and relentless assistance and appreciation was indispensable. ...

Nomenclature is important. Despite the efforts, in landslide recognition and mapping (Varnes, 1978; Cruden, 1991; Cruden and Varnes, 1996; WP/WLI, 1990, 1993, 1995), susceptibility and hazard assessment (Varnes and IAEG Commission on landslides and other mass movements, 1984; Aleotti and Chowdhury, 1999; Guzzetti et al., 1999; Partnership for Reducing Landslide Risk, 2004) and risk evaluation (Varnes and IAEG Commission on landslides and other mass movements, 1984; Einstein, 1988; 1997; Cruden and Fell, 1997; ISSMGE TC32, 2004; Partnership for Reducing Landslide Risk, 2004; Vandine et al., 2004; Wise et al., 2004; Glade et al., 2005), confusion exists on the use and application of many terms. This often results in difficulty in comparing the results of different investigators. The simplified glossary presented in this chapter does not have the ambition of solving the problem. In the following, some of the most important terms or expressions used in this work are listed. For each term a short explanation is provided and, where appropriate, reference is made to the relevant literature. Meaning of some of the language used in this work may not be the same as that found in the literature. ...

I have a passion for reading. I spent time in public, sheared and personal libraries, searching the literature and the Internet, exchanging papers, reports, maps and information with colleagues and friends, photocopying, printing, reading and digesting information on landslide recognition, mapping, susceptibility and hazard assessment, risk evaluation and mitigation, planning and policy making, and related topics. The following list of references is the result of this work. ...

Variables used in the text ...

List of Figures and Tables shown in the dissertation ...

List of acronyms used in the dissertation ...