Prediction of landslide susceptibility using rare events logistic regression: A case-study in the Flemish Ardennes (Belgium)

In this article a statistical multivariate method, i.e., rare events logistic regression, is evaluated for the creation of a landslide susceptibility map in a 200 km² study area of the Flemish Ardennes (Belgium). The methodology is based on the hypothesis that future landslides will have the same causal factors as the landslides initiated in the past. The information on the past landslides comes from a landslide inventory map obtained by detailed field surveys and by the analysis of LIDAR (Light Detection and Ranging)-derived hillshade maps. Information on the causal factors (e.g., slope gradient, aspect, lithology, and soil drainage) was extracted from digital elevation models derived from LIDAR and from topographical, lithological and soil maps. In landslide-affected areas, however, we did not use the present-day hillslope gradient. In order to reflect the hillslope condition prior to landsliding, the pre-landslide hillslope was reconstructed and its gradient was used in the analysis. Because of their limited spatial occurrence, the landslides in the study area can be regarded as “rare events”. Rare events logistic regression differs from ordinary logistic regression because it takes into account the low proportion of 1s (landslides) to 0s (no landslides) in the study area by incorporating three correction measures: the endogenous stratified sampling of the dataset, the prior correction of the intercept and the correction of the probabilities to include the estimation uncertainty. For the study area, significant model results were obtained, with pre-landslide hillslope gradient and three different clayey lithologies being important predictor variables. Receiver Operating Characteristic (ROC) curves and the Kappa index were used to validate the model. Both show a good agreement between the observed and predicted values of the validation dataset. Based on a qualified judgement, the created landslide susceptibility map was classified into four classes, i.e., very high, high, moderate and low susceptibility. If interpreted correctly, this classified susceptibility map is an important tool for the delineation of zones where prevention measures are needed and human interference should be limited in order to avoid property damage due to landslides.     

Probabilistic landslide hazard assessment at the basin scale

We propose a probabilistic model to determine landslide hazard at the basin scale. The model predicts where landslides will occur, how frequently they will occur, and how large they will be. We test the model in the Staffora River basin, in the northern Apennines, Italy. For the study area, we prepare a multi-temporal inventory map through the interpretation of multiple sets of aerial photographs taken between 1955 and 1999. We partition the basin into 2243 geo-morpho-hydrological units, and obtain the probability of spatial occurrence of landslides by discriminant analysis of thematic variables, including morphological, lithological, structural and land use. For each mapping unit, we obtain the landslide recurrence by dividing the total number of landslide events inventoried in the unit by the time span of the investigated period. Assuming that landslide recurrence will remain the same in the future, and adopting a Poisson probability model, we determine the exceedance probability of having one or more landslides in each mapping unit, for different periods. We obtain the probability of landslide size by analysing the frequency–area statistics of landslides, obtained from the multi-temporal inventory map. Assuming independence, we obtain a quantitative estimate of landslide hazard for each mapping unit as the joint probability of landslide size, of landslide temporal occurrence and of landslide spatial occurrence.     

Mapping landslide susceptibility from small datasets: A case study in the Pays de Herve (E Belgium)

A landslide susceptibility map is proposed for the Pays de Herve (E Belgium), where large landslides affect Cretaceous clay outcrop areas. Based on a Bayesian approach, this GIS-supported probabilistic map identifies the areas most susceptible to deep landslides. The database is comprised of the source areas of ten pre-existing landslides (i.e. a sample of 154 grid cells) and of six environmental data layers, namely lithology, proximity to active faults, slope angle and aspect, elevation and distance to the nearest valley-floor. A 30-m-resolution DEM from the Belgian National Geographical Institute is used for the analysis. Owing to the small size of the sample, a special cross-validation procedure of the susceptibility map is performed, which uses in an iterative way each of the landslides to test the predictive power of the map derived from the other landslides. Four different sets of variables are used to produce four susceptibility maps, whose prediction curves are compared. While the prediction rates associated with the models not involving the “proximity to active fault” criterion are comparable to those of the models considering this variable, strong weaknesses inherent in the fault data on which the latter rely suggest that the final susceptibility map should be based on a model that excludes any reference to fault. This highlights the difference between a triggering factor and determining factors, and in the same time broadens the scope of the produced map. A single reactivated slide is also used to test the possibility of predicting future reactivation of existing landslides in the area. Finally, the need for geomorphological control over the mathematical treatment is underlined in order to obtain realistic prediction maps.     

Spatial patterns of old, deep-seated landslides: A case-study in the northern Ethiopian highlands

During the last decade, slope failures were reported in a 500 km² study area in the Geba–Werei catchment, northern Ethiopia, a region where landslides were not considered an important hazard before. Field observations, however, revealed that many of the failures were actually reactivations of old deep-seated landslides after land use changes. Therefore, this study was conducted (1) to explore the importance of environmental factors controlling landslide occurrence and (2) to estimate future landslide susceptibility. A landslide inventory map of the study area derived from aerial photograph interpretation and field checks shows the location of 57 landslides and six zones with multiple landslides, mainly complex slides and debris flows. In total 14.8% of the area is affected by an old landslide. For the landslide susceptibility modelling, weights of evidence (WofE), was applied and five different models were produced. After comparison of the models and spatial validation using Receiver Operating Characteristic curves and Kappa values, a model combining data on elevation, hillslope gradient, aspect, geology and distance to faults was selected. This model confirmed our hypothesis that deep-seated landslides are located on hillslopes with a moderate slope gradient (i.e. 5°–13°). The depletion areas are expected on and along the border of plateaus where weathered basalts rich in smectite clays are found, and the landslide debris is expected to accumulate on the Amba Aradam sandstone and upper Antalo limestone. As future landslides are believed to occur on inherently unstable hillslopes similar to those where deep-seated landslides occurred, the classified landslide susceptibility map allows delineating zones where human interventions decreasing slope stability might cause slope failures. The results obtained demonstrate that the applied methodology could be used in similar areas where information on the location of landslides is essential for present-day hazard analysis.     

Predictive modelling of rainfall-induced landslide hazard in the Lesser Himalaya of Nepal based on weights-of-evidence

Landslide hazard mapping is a fundamental tool for disaster management activities in mountainous terrains. The main purpose of this study is to evaluate the predictive power of weights-of-evidence modelling in landslide hazard assessment in the Lesser Himalaya of Nepal. The modelling was performed within a geographical information system (GIS), to derive a landslide hazard map of the south-western marginal hills of the Kathmandu Valley. Thematic maps representing various factors (e.g., slope, aspect, relief, flow accumulation, distance to drainage, soil depth, engineering soil type, landuse, geology, distance to road and extreme one-day rainfall) that are related to landslide activity were generated, using field data and GIS techniques, at a scale of 1:10,000. Landslide events of the 1970s, 1980s, and 1990s were used to assess the Bayesian probability of landslides in each cell unit with respect to the causative factors. To assess the accuracy of the resulting landslide hazard map, it was correlated with a map of landslides triggered by the 2002 extreme rainfall events. The accuracy of the map was evaluated by various techniques, including the area under the curve, success rate and prediction rate. The resulting landslide hazard value calculated from the old landslide data showed a prediction accuracy of > 80%. The analysis suggests that geomorphological and human-related factors play significant roles in determining the probability value, while geological factors play only minor roles. Finally, after the rectification of the landslide hazard values of the new landslides using those of the old landslides, a landslide hazard map with > 88% prediction accuracy was prepared. The methodology appears to have extensive applicability to the Lesser Himalaya of Nepal, with the limitation that the model's performance is contingent on the availability of data from past landslides.     

Terrain Pattern Recognition and Spatial Decision Making for Regional Slope Stability Studies

A terrain partition scheme is presented that allows the identification of regions with high landslide risk in natural terrain zones on the basis of geomorphometric criteria from moderate resolution DEMs. The key factor being the terrain segmentation to aspect regions (regions formed by points preserving the same aspect direction) instead of using an artificial regular-grid terrain partition scheme. The study area is in western Greece (NW Peloponnesus) whereas a moderate resolution digital elevation model with spacing 75 m is used. Landslide inventory analysis and knowledge conceptualization identified that the landslide susceptibility of a particular aspect region is high, if the mean elevation is low and the mean gradient is high. Each aspect region was parametrically represented on the basis of its mean gradient and elevation. The domain of each parameter was divided to seven slices (classes) on the basis of the observed density. Subsequent knowledge based mapping identified aspect regions with high landslide susceptibility for the following spatial rule: (a) “mean slope in class 6 or 7” and (b) “mean elevation in class 1 to 5”. Alternatively the rule is expressed as mean slope to be equal or greater than 15^∘ whereas mean elevation to be in the range 0 to 750 m. These identified zones correspond to regions where historical landslides occurred (populated coastal areas in the North) as well as to south regions (natural terrain zone) where no landslide record is available, because of the limitations posed by the natural terrain landslide mapping program in Greece. The presented terrain segmentation technique combined to the spatial decision-making process, provided both an object framework for integrating geomorphometric parameters and a method for landslide risk analysis in natural terrain zones.     

Landslide inventory in a rugged forested watershed: a comparison between air-photo and field survey data

Landslide inventories are routinely compiled by means of aerial photo interpretation (API). When examining photo pairs, the forest canopy (notably in old-growth forest) hides a population of “not visible” landslides. In the present study, we attempt to estimate how important is the contribution of landslides not detectable from aerial photographs to the global mass of sediment production from mass failures on forested terrain of the Capilano basin, coastal British Columbia. API was coupled with intensive fieldwork for identification and measurement of all landslides. A 30-year framework was adopted. We show that “not visible” landslides can represent up to 85% of the total number of failures and account for 30% of the volume of debris mobilised. Such percentages display high sub-basin variability with rates of sediment production varying by one order of magnitude between two sub-basins of the study area. This is explained qualitatively by GIS-based analysis of slope frequency distributions, drainage density, and spatial distribution of surficial materials. Such observations find further support in the definitions of transport-limited and supply-limited basins. As a practical consideration to land managers, we envisage that supplementary fieldwork for landslide identification is mandatory in transport-limited systems only. Fieldwork has demonstrated that gully-related failures have a greater importance than one could expect from API.     

Landslide susceptibility mapping based on Support Vector Machine: A case study on natural slopes of Hong Kong, China

The Support Vector Machine (SVM) is an increasingly popular learning procedure based on statistical learning theory, and involves a training phase in which the model is trained by a training dataset of associated input and target output values. The trained model is then used to evaluate a separate set of testing data. There are two main ideas underlying the SVM for discriminant-type problems. The first is an optimum linear separating hyperplane that separates the data patterns. The second is the use of kernel functions to convert the original non-linear data patterns into the format that is linearly separable in a high-dimensional feature space. In this paper, an overview of the SVM, both one-class and two-class SVM methods, is first presented followed by its use in landslide susceptibility mapping. A study area was selected from the natural terrain of Hong Kong, and slope angle, slope aspect, elevation, profile curvature of slope, lithology, vegetation cover and topographic wetness index (TWI) were used as environmental parameters which influence the occurrence of landslides. One-class and two-class SVM models were trained and then used to map landslide susceptibility respectively. The resulting susceptibility maps obtained by the methods were compared to that obtained by the logistic regression (LR) method. It is concluded that two-class SVM possesses better prediction efficiency than logistic regression and one-class SVM. However, one-class SVM, which only requires failed cases, has an advantage over the other two methods as only “failed” case information is usually available in landslide susceptibility mapping.     

GIS支持下三峡库区秭归县滑坡灾害空间预测

基于GIS空间分析和统计模型相结合进行区域评价与空间预测是滑坡灾害研究的重要方向之一。以三峡库区秭归县为研究区,选择坡度、坡向、边坡结构、工程岩组、排水系统、土地利用和公路开挖作为评价因子。为提高模型的预测精度、可信度和推广能力,利用窗口采样规则降低训练样本之间的空间相关性。建立Logistic回归模型,对滑坡灾害与评价因子进行定量相关性分析。计算研究区滑坡灾害易发性指数,对其进行聚类分析,绘制滑坡易发性分区图,其中高、中易发区占整个研究区面积的38.9%,主要分布在人类工程活动频繁和靠近排水系统的区域。经过验证,该模型的预测精度达到77.57%。     

Comparison of landslide susceptibility based on a decision-tree model and actual landslide occurrence: The Akaishi Mountains, Japan

This paper proposes a statistical decision-tree model to analyze landslide susceptibility in a wide area of the Akaishi Mountains, Japan. The objectives of this study were to validate the decision-tree model by comparing landslide susceptibility and actual landslide occurrence, and to reveal the relationships among landslide occurrence, topography, and geology. Landslide susceptibility was examined through ensemble learning with a decision tree. Decision trees are advantageous in that estimation processes and order of important explanatory variables are explicitly represented by the tree structures. Topographic characteristics (elevation, slope angle, profile curvature, plan curvature, and dissection and undissection height) and geological data were used as the explanatory variables. These topographic characteristics were calculated from digital elevation models (DEMs). The objective variables were landslide occurrence and reactivation data between 1992 and 2002 that were depicted by satellite image analysis. Landslide susceptibility was validated by comparing actual data on landslides that occurred and reactivated after the model was constructed (between 2002 and 2004).This study revealed that, from 2002 to 2004, landslides tended to occur and reactivate in catchments with high landslide susceptibility. The landslide susceptibility map thus depicts the actual landslide occurrence and reactivation in the Akaishi Mountains. This result indicates that the decision-tree model has appropriate accuracy for estimating the probabilities of future landslides. The tree structure indicates that landslides occurred and reactivated frequently in the catchments that had an average slope angle exceeding ca. 29° and a mode of slope angle exceeding 33°, which agree well with previous studies. A decision tree also quantitatively expresses important explanatory variables at the higher order of the tree structure.     

Geomorphic hazard assessment of landslide dams in South Westland, New Zealand: fundamental problems and approaches

Quantitative assessments of landslide hazard usually employ empirical, heuristic, deterministic, or statistical methods to derive estimates of magnitude–frequency distributions of landsliding. The formation and failure of landslide dams are common geomorphic processes in mountain regions throughout the world, causing a series of consequential off-site hazards such as catastrophic outburst floods, debris flows, backwater ponding, up- and downstream aggradation, and channel instability.Conceptual and methodological problems of quantifying geomorphic hazard from landslide dams result from (a) aspects of defining “landslide-dam magnitude”, (b) scaling effects, i.e. the geomorphic long-range and long-term implications of river blockage, and (c) paucity of empirical data. Geomorphic hazard from a landslide dam-break flood on the basis of conditional probabilities is being analysed for the alpine South Westland region of New Zealand, where formation and failure of landslide dams is frequent. Quantification of the annual probability of landsliding and subsequent dam formation in the area is limited by historical and only partially representative empirical data on slope instability. Since landslide-dam stability is a major control governing the potential of catastrophic outburst flooding, the ensuing hazard is best assessed on a recurring basis. GIS-based modelling of virtual landslide dams is a simple and cost-effective approach to approximate site-specific landslide dam and lake dimensions, reservoir infill times, and scaled magnitude of potential outburst floods. Although crude, these order-of-magnitude results provide information critical to natural hazard planning, mitigation, or emergency management decisions.     

Probabilistic landslide risk analysis considering direct costs in the area north of Lisbon (Portugal)

The purpose of the present study is the analysis of landslide risk for roads and buildings in a small test site (20 km²) in the area north of Lisbon (Portugal). For this purpose, an evaluation is performed integrating into a GIS information obtained from multiple sources: (i) landslide hazard; (ii) elements at risk; and (iii) vulnerability. Landslide hazard is assessed on a probabilistic basis for three different types of slope movement (shallow translational slides, translational slides and rotational slides), based on some assumptions such as: (i) the likelihood of future landslide occurrence can be measured through statistical relationships between past landslide distribution and specified spatial data sets considered as landslide predisposing factors; and (ii) the rainfall combination (amount–duration) responsible for past slope instability within the test site will produce the same effects (i.e. same type of landslides and similar total affected area), each time they occur in the future. When the return period of rainfall triggering events is known, different scenarios can be modelled, each one ascribed to a specific return period. Therefore, landslide hazard is quantitatively assessed on a raster basis, and is expressed as the probability for each pixel (25 m²) to be affected by a future landslide, considering a rainfall triggering scenario with a specific return period. Elements at risk within the test site include 2561 buildings and roads amounting to 169 km. Values attributed to elements at risk were defined considering reconstruction costs, following the guidelines of the Portuguese Insurance Institute. Vulnerability is considered as the degree of loss to a given element resulting from the occurrence of a landslide of a given magnitude. Vulnerability depends not only on structural properties of exposed elements, but also on the type of process, and its magnitude; i.e., vulnerability cannot be defined in absolute terms, but only with respect to a specific process (e.g. vulnerability to shallow translational slides). Therefore, vulnerability was classified for the three landslide groups considered on hazard assessment, taking into account: (i) landslide magnitude (mean depth, volume, velocity); (ii) damage levels produced by past landslide events in the study area; and (iii) literature. Finally, a landslide risk analysis considering direct costs was made in an automatic way crossing the following three layers: (i) Probabilistic hazard map for a landslide type Z, considering a particular rainfall triggering scenario whose return period is known; (ii) Vulnerability map (values from 0 to 1) of the exposed elements to landslide type Z; and (iii) Value map of the exposed elements, considering reconstruction costs.     

GIS and ANN model for landslide susceptibility mapping

Landslide hazard is as the probability of occurrence of a potentially damaging landslide phenomenon within specified period of time and within a given area. The susceptibility map provides the relative spatial probability of landslides occurrence. A study is presented of the application of GIS and artificial neural network model to landslide susceptibility mapping, with particular reference to landslides on natural terrain in this paper. The method has been applied to Lantau Island, the largest outlying island within the territory of Hong Kong. A three-level neural network model was constructed and trained by the back-propagate algorithm in the geographical database of the study area. The data in the database includes digital elevation modal and its derivatives, landslides distribution and their attributes, superficial geological maps, vegetation cover, the raingauges distribution and their 14 years 5-minute observation. Based on field inspection and analysis of correlation between terrain variables and landslides frequency, lithology, vegetation cover, slope gradient, slope aspect, slope curvature, elevation, the characteristic value, the rainstorms corresponding to the landslide, and distance to drainage line are considered to be related to landslide susceptibility in this study. The artificial neural network is then coupled with the ArcView3.2 GIS software to produce the landslide susceptibility map, which classifies the susceptibility into three levels: low, moderate, and high. The results from this study indicate that GIS coupled with artificial neural network model is a flexible and powerful approach to identify the spatial probability of hazards.     

GIS and ANN model for landslide susceptibility mapping

1 IntroductionThe population growth and the expansion of settlements and life-lines over hazardous areas exert increasingly great impact of natural disasters both in the developed and developing countries. In many countries, the economic losses and casualties due to landslides are greater than commonly recognized and generate a yearly loss of property larger than that from any other natural disasters, including earthquakes, floods and windstorms. Landslides in mountainous terrain often occur a…     

On the spatial relationship between landslides and causative factors on Lantau Island, Hong Kong

This paper presents a statistical approach to study the spatial relationship between landslides and their causative factors at the regional level. The approach is based on digital databases, and incorporates such methods as statistics, spatial pattern analysis, and interactive mapping. Firstly, the authors propose an object-oriented conceptual model for describing a landslide event, and a combined database of landslides and environmental factors is constructed by integrating the various databases within such a conceptual framework. The statistical histogram, spatial overlay, and dynamic mapping methods are linked together to interactively evaluate the spatial pattern of the relationship between landslides and their causative factors. A case study of an extreme event in 1993 on Lantau Island indicates that rainfall intensity and the migration of the center of the rainstorm greatly influence the occurrence of landslides on Lantau Island. A regional difference in the relationship between landslides and topography is identified. Most of the landslides in the middle and western parts of the island occurred on slopes with slope angles of 25–35°, while in the eastern part, the corresponding range is 30–35°. Overlaying landslide data with land cover reveals that a large number of landslides occurred in the bareland and shrub-covered area, and in the transition zones between different vegetation types. The proposed approach can be used not only to analyze the general characteristics of such a relationship, but also to depict its spatial distribution and variation, thereby providing a sound basis for regional landslide prediction.     

Spatial pattern and influencing factors of landslide casualty events

Analysis of casualties due to landslides from 2000 to 2012 revealed that their spatial pattern was affected by terrain and other natural environmental factors, which resulted in a higher distribution of landslide casualty events in southern China than in northern China. Hotspots of landslide-generated casualties were in the western Sichuan mountainous area and Yunnan-Guizhou Plateau region, southeast hilly area, northern part of the loess hilly area, and Tianshan and Qilian Mountains. However, local distribution patterns indicated that landslide casualty events were also influenced by economic activity factors. To quantitatively analyse the influence of natural environment and human-economic activity factors, the Probability Model for Landslide Casualty Events in China (LCEC) was built based on logistic regression analysis. The results showed that relative relief, GDP growth rate, mean annual precipitation, fault zones, and population density were positively correlated with casualties caused by landslides. Notably, GDP growth rate ranked only second to relative relief as the primary factors in the probability of casualties due to landslides. The occurrence probability of a landslide casualty event increased 2.706 times with a GDP growth rate increase of 2.72%. In contrast, vegetation coverage was negatively correlated with casualties caused by landslides. The LCEC model was then applied to calculate the occurrence probability of landslide casualty events for each county in China. The results showed that there are 27 counties with high occurrence probability but zero casualty events. The 27 counties were divided into three categories: poverty-stricken counties, mineral-rich counties, and real-estate overexploited counties; these are key areas that should be emphasized in reducing landslide risk.     

中国地质灾害伤亡事件的空间格局及影响因素（英文）

Analysis of casualties due to landslides from 2000 to 2012 revealed that their spatial pattern was affected by terrain and other natural environmental factors, which resulted in a higher distribution of landslide casualty events in southern China than in northern China. Hotspots of landslide-generated casualties were in the western Sichuan mountainous area and Yunnan-Guizhou Plateau region, southeast hilly area, northern part of the loess hilly area, and Tianshan and Qilian Mountains. However, local distribution patterns indicated that landslide casualty events were also influenced by economic activity factors. To quantitatively analyse the influence of natural environment and human-economic activity factors, the Probability Model for Landslide Casualty Events in China(LCEC) was built based on logistic regression analysis. The results showed that relative relief, GDP growth rate, mean annual precipitation, fault zones, and population density were positively correlated with casualties caused by landslides. Notably, GDP growth rate ranked only second to relative relief as the primary factors in the probability of casualties due to landslides. The occurrence probability of a landslide casualty event increased 2.706 times with a GDP growth rate increase of 2.72%. In contrast, vegetation coverage was negatively correlated with casualties caused by landslides. The LCEC model was then applied to calculate the occurrence probability of landslide casualty events for each county in China. The results showed that there are 27 counties with high occurrence probability but zero casualty events. The 27 counties were divided into three categories: poverty-stricken counties, mineral-rich counties, and real-estate overexploited counties; these are key areas that should be emphasized in reducing landslide risk.     

基于GIS的澜沧江下游区滑坡灾害危险性分析

澜沧江流域是中国西南地区滑坡灾害较为严重的地区。对澜沧江下游区滑坡灾害及其控制因素分析,建立基于G IS的滑坡灾害危险性评价模型,实现澜沧江下游区滑坡危险性区划,为该区滑坡灾害防治和生态环境保护等提供重要决策依据。     

基于空间密度函数的哈尼梯田世界遗产地滑坡时空格局变化

为探究哈尼梯田世界文化景观遗产地核心区滑坡灾害时空分布规律，以Google Earth 0.55 m分辨率的2005、2009、2015年3期遥感影像为基础，结合实地走访调查，建立滑坡数据库，在ArcGIS 10.2平台上计算滑坡点的最邻近指数、K函数曲线及密度分布。结果显示：1）哈尼梯田遗产核心区2005、2009、2015年的滑坡数量分别为184、337和285个，对应最邻近指数为0.556、0.603、0.628；最显著聚集的空间尺度为1 000 m，从聚集向离散分布转变的空间尺度阈值分别为2.9、3.9、3.6 km。2）3个年份滑坡点高密度区占比逐渐增加（2.3%→5.8%→8.3%），中密度区占比亦逐渐增大（15.7%→21.8%→27.9%），低密度区占比逐渐减小（82.0%→72.5%→66.8%）。3）需要重点防范滑坡灾害风险的区域为森林区的西段和东段，村寨区的多依树、硐浦、勐品、水卜龙等地，以及阿勐控河和碧猛河流域内的梯田区。综上，研究区2005－2015年滑坡空间格局发生了显著变化，随着人类活动对地表景观干预程度不断加大，滑坡灾害风险增加了更多的不确定性。     

Quantitative landslide risk assessment and mapping on the basis of recent occurrences

A quantitative procedure for mapping landslide risk is developed from considerations of hazard, vulnerability and valuation of exposed elements. The approach based on former work by the authors, is applied in the Bajo Deba area (northern Spain) where a detailed study of landslide occurrence and damage in the recent past (last 50 years) was carried out. Analyses and mapping are implemented in a Geographic Information System (GIS).The method is based on a susceptibility model developed previously from statistical relationships between past landslides and terrain parameters related to instability. Extrapolations based on past landslide behaviour were used to calculate failure frequency for the next 50 years. A detailed inventory of direct damage due to landslides during the study period was carried out and the main elements at risk in the area identified and mapped. Past direct (monetary) losses per type of element were estimated and expressed as an average ‘specific loss’ for events of a given magnitude (corresponding to a specified scenario). Vulnerability was assessed by comparing losses with the actual value of the elements affected and expressed as a fraction of that value (0–1).From hazard, vulnerability and monetary value, risk was computed for each element considered. Direct risk maps (€/pixel/year) were obtained and indirect losses from the disruption of economic activities due to landslides assessed. The final result is a risk map and table combining all losses per pixel for a 50-year period. Total monetary value at risk for the Bajo Deba area in the next 50 years is about 2.4 × 10⁶ Euros.