1. INTRODUCTIONLandslides are a high-risk natural phenomenon that occurs all around the world. Globally landslide has caused approximately 1000 deaths per year (Lee and Pradhan 2007). In addition to these causalities, these landslides poses serious threat to structures that supports transportation, settlements and tourism. Basically, a landslide is a downward movement of soil, rock or debris occurring either in a curved or planar surface of rupture as a coherent mass with little or no internal deformation. Landslides can effect a man-made structure when they are in direct or indirect path of a landslide. Landslide damage to lifelines such as sewer line, water line or electric line can cause disruption of services to an entire region as a whole. Commercial structures if affected can cause damage to the building itself and also to the whole community due to interruption of business. One of the largest consequences due to landslides can be to the transportation infrastructures. Landslides commonly occurs on cut slopes or embankments alongside roads and railways disrupting the commute. In addition to the potential fatalities, delays in the travel time due to failure of slopes can bring about monetary losses as well. Various factors contribute to the occurrence of landslides. It can be a). Natural & b). Man-Made. Natural category of causes include three main triggering factors i). Water, ii). Seismic Activity and iii). Volcanic Activity which can either occur in single or in combination. Effects of all these causes can widely vary based on the slope geometry, soil type, whether there are people or structures on the affected area etc. Man-Made causes are majorly due to population expanding onto new land and creating new neighbourhood and towns. Steepening of slopes for development, changes in loading patterns, disturbing or modifying drainage patterns, removing vegetation etc. are some of the reasons that can act as a triggering factor for landslides. Saturation by water is a primary cause of landslides. It can occur in the form of precipitation, snow melt, change in ground water levels etc. Chiefly variation in precipitation due to climate change can control or influence the occurrence of majority of landslides (Ciabatta, Camici et al. 2016). A study focused on finding the major causes for landslides (Kazmi, Qasim et al. 2017) suggests that rainfall contributes majority (62%) as the triggering factor for landslides. Effect of rainfall is even more pronounced in tropical and sub-tropical countries. Storm induced landslides are common in many tropical or sub-tropical countries like Singapore, Malaysia, Taiwan, Japan, Hong Kong etc. (Zhang, Zhang et al. 2011). Climate change due to global warming is expected to lead to a greater frequency and magnitude of heavy precipitation. For example, Dore (2005) infers that due to climate change, northern hemisphere will receive a larger amount of precipitation while in the equatorial regions frequency is likely to be effected. Considering the fact that precipitation is most common trigger to the landslide activity, it’s not very surprising that there is a strong theoretical basis for increased landslide activity as a result of climate-change.Earthworks slopes (embankments and cutting) forms a major portion of any transportation networks. As described above, a landslide that occurs on cut slopes or embankments alongside roads and railways can bring about major delays and disruption to network operation. In an engineering perspective of sustainability considering the effects of climate change in the slope stability, there can be potential social and economic risks which need to be addressed. Assessing this risk and adapting the infrastructure to a changing climate scenario requires knowledge about resilience of these structures to projected climate change. 1.1 Landslide Risk AnalysisQuantifying the increased landslide activity and risk associated with it due to climate change warrants a methodical approach to be implemented. Quantitative Risk Assessment (QRA) for engineered slopes and landslides is a framework described in terms of 4 main analysis steps (Fig. 1). This is achieved by breaking up the whole process into smaller process of different disciplines. Process are formulated such that they all are independent and sequential in nature so that each of the process can be handled by the experts in each field more clearly and rigorously. Four Major process in QRA are Hazard Analysis, Risk Analysis, Risk Evaluation and Risk Mitigation. Hazard analysis includes landslide characterization and calculating frequency associated with landslides. Risk analysis includes both hazard analysis and consequence analysis. Risk is defined as measure of the probability of an adverse effect to life, health and environment (Fell, Ho et al. 2005). Quantitatively, Risk = Hazard x Consequence.Consequence can also be termed as potential worth of loss. Loss can be either in terms of property or persons. Consequence analysis involves identifying these elements of potential loss and quantifying its probability or vulnerability in terms of conditional probability with respect to hazard frequency. Hazard is the probability that a particular danger occurs within a given period of time (Fell, Ho et al. 2005). Occurrence of a landslide due to failure of slope can be considered as a hazard event and probability of that particular slope to fail in a given period of time (e.g. a year) is the hazard associated with that event. Hazard analysis forms the basis of risk analysis using the QRA framework. Figure 1. Schematic representation of Quantitative Risk Assessment Process (adapted from Fell, Ho et al. (2005))1.2 Hazard AnalysisHazard analysis includes the process of identification and characterisation of the potential landslides to evaluate their corresponding frequency of occurrence. Several methods can be employed to calculate frequency of landslides. Some of the methods are empirical in which historical or geomorphological evidence is analysed to form conclusion regarding future performance. Empirical methods in most cases provide qualitative results in terms of ill-defined subjective terms. Quantitative results can be obtained by physically or numerically modelling the slope system along with varied levels of information regarding slope geometry, soil shear strength etc. Quantitative results are more preferred as it provides uniform outcomes in quantified terms (Fell, Ho et al. 2005). Quantitative results also helps in comparing risks with quantitative acceptance criteria. Owing to a wide range of uncertainty simulating field conditions in modelling, researchers and practising engineers often rely on probabilistic framework to assess the performance of a given structure. Application of probabilistic approach along with physical or numerical modelling have the advantage of being physically accurate and able to give reliable quantitative results. To predict the potential risks due to slopes failure associated with climate change, probabilistic methods that can represent the performance of a slope as a function of external climatic variables is beneficial as it will allow flexibility in the next stage of risk analysis. Performance of slopes can be characterized through fragility curves which represents the slope performance as a function of climatic variables like rainfall. 1.3 Fragility Curves.Fragility curves generally describe conditional probabilities of the state of structure to have exceeded a demand condition. Demand condition is the expected level of performance from a structure below which structure is deemed to be failed to meet the demand. Thus, performance of a structure based on a particular demand can be represented probabilistically using a variable called probability of failure. Probability of failure is defined as the probability of structure being not able to meet the performance criteria. Fragility curves involves calculating relative probability of failures of a structure via conventional probabilistic analysis over a range of loading conditions. Schultz, Gouldby et al. (2010) states that fragility curves provides a richer, more comprehensive perspective on the reliabilities of geostructures, because they are functions rather than points. Fragility curves is one of the important statistical tool employed to represent the hazard frequency as it has the advantage of representing the hazard over a range of external variables which is beneficial to allow flexibility while assessing the risk in the later stage. Despite the extensive application of fragility curves in the risk analysis slopes subjected to seismic shaking, use of fragility curves to assess the risk of slopes subjected to rainfall has been curtailed. A comprehensive approach to developing fragility curves for slopes subjected to rainfall will shed the light on the efficiency of fragility curves as a hazard analysis tool. This report is divided into four main chapters. Chapter one is an introductory chapter explaining the motivation behind this research work. It details about the effect of climate change in the performance of slopes, basic framework of landslide risk analysis and the concept of fragility curves. Chapter two presents a selective review of literatures published with particular importance to fragility curves and its application. Chapter three identifies the research gap and state the aim & objectives of this research work. Chapter four details some preliminary work and associated results. It also gives an outline of methodology that will be followed to achieve the objectives stated.2. LITERATURE REVIEWFragility curves are being increasingly used as a common component of risk assessments. Concept of fragility curves is applied to several structural/geotechnical systems by many researchers. Recently, fragility curves are being used for flood risk assessment as well. Different approaches for the development of fragility curves can be classified into 4 main categories: judgemental, empirical, analytical and hybrid (Schultz, Gouldby et al. 2010). Judgmental approach is based on engineering judgment and expert opinion. This method is highly susceptible to bias and rarely used due to difference in individual experience and influence of specific location in which the experts have experience. Empirical approach is based on previous observational data on the performance of structures under variety of loads. It can be accomplished through controlled experiments or may be collected as opportunity comes in which case it is uncontrolled. Empirical methods can be used for non-structural systems, electrical and mechanical parts as these are easy to test and the specimen can be loaded up to failure. Sometimes natural hazard events can also yield sufficient number of data points to accurately calculate the fragility curves. Shinozuka, Feng et al. (2000) and Tanaka, Kameda et al. (2000) have developed fragility curves from bridge inspection data collected at bridges following the Hoboken Nambu earthquake. Casciati, Cimellaro et al. (2008) noted that empirical fragility curves are more-or-less specific to observed structures and it cannot be used to model the reliability of other structures. Analytical fragility curves are based on structural models that characterize the performance of the structures. The performance of the structure is represented as a function of several random variables defining the capacity and demand placed on the structure. The fragility curve is constructed by calculating a probability of failure under load ranging from those at which failure is highly unlikely to those at which failure is almost certain. Hybrid methods try to absorb merits of different methods to accurately predict the fragility curves. While each approach has its own merits and drawbacks, analytical approaches are most commonly used due to its flexibility and mathematical approachability. In the context of risk assessment also, analytical methods based on structural models are more preferred as it provides uniform outcomes in quantified terms and helps in comparing the risk with quantitative acceptance criteria. Methodology for developing fragility curves through analytical approach is well studied with special reference nuclear structures, concrete structures like dam & bridges and retaining structures as well. Bradley, Cubrinovski et al. (2010) applied assessed the seismic performance and loss of a bridge-foundation soil system. A finite element method incorporating advanced models for bridge and soil elements is used to predict the earthquake response of the system. In all nine EDPs are identified through a number of deterministic analyses. A seismic demand hazard curve is developed for all the parameters by combining the seismic hazard curve and seismic response analysis. Argyroudis, Kaynia et al. (2013) developed fragility curves for different damage states of a cantilever retaining wall. A 2D finite element nonlinear model with elastoplastic soil behavior is used to simulate the response of cantilever bridge abutments to varying levels of seismic intensity. Damage state is defined based on the EDP i.e., vertical settlement of the backfill immediately behind the wall. Effect of soil condition and ground motion characteristics on the global soil and structure response is studied by considering different soil profiles and seismic input motions Bradley, Cubrinovski et al. (2010) used fragility curves of dike sections to appropriately integrate geostatic and geohydraulic dike characteristics into operational flood management system. A case study of River Emscher (Germany) was selected. Within the study, fragility curves for 33 different dike cross sections on the lower reach of the Emscher were developed. Monte Carlo simulations were used to find the probability of failure. The stress variable in which the failure is conditioned was considered as water level (h). Segregated fragility curves for 8 different failure modes and a total fragility curve using 3 approaches are calculated. This facilitated an extended reliability assessment of the dam with quick access and simple interpretability of relevant information.Ebeling, Fong et al. (2012) employed numerical methods to develop fragility curves for concrete dams embedded in rocks considering the sliding mode of failure. They used Latin Hypercube sampling along with a PC-based software environment, GDLAD_Foundation to calculate fragility of sliding for a concreted dam. Vorogushyn, Merz et al. (2009) described development of fragility curves for the slope stability failure caused by seepage through earthen dikes using Monte Carlo simulation (MCS). Reliability functions were formulated based on two major breach mechanisms in the dike. Reliability functions were formed based on the review of physically-based and empirical process descriptions which when analysed on the framework of MCS led to the development of fragility curves.