Skip to main content

Living with landslide risk in Europe: Assessment, effects of global change, and risk management strategies

Final Report Summary - SAFELAND (Living with landslide risk in Europe: Assessment, effects of global change, and risk management strategies)

SAFELAND is a large, integrating research project under the European Commission (EC)'s Seventh Framework Programme (FP7). The project started on 1 May 2009 and ended on 30 April 2012. It involved 27 partners from 12 European countries, and had international collaborators and advisers from China, India, United States (US), Japan and Hong Kong. SAFELAND also involved 25 end-users from 11 countries. SAFELAND was coordinated by the International Centre for Geohazards (ICG) at Norwegian Geotechnical Institute in Norway.

Main results achieved in SAFELAND include:

- various guidelines related to landslide triggering processes and run-out modelling;
- development and testing of several empirical methods for predicting the characteristics of threshold rainfall events for triggering of precipitation-induced landslides, and development of an empirical model for assessing the changes in landslide frequency (hazard) as a function of changes in the demography and population density;
- guidelines for landslide susceptibility, hazard and risk assessment and zoning;
- new methodologies for physical and societal vulnerability assessment;
- identification of landslide hazard and risk hotspots for Europe;
- different regional and local climate model simulations over selected regions of Europe at spatial resolutions of 10x10 km and 2.8x2.8 km;
- guidelines for use of remote sensing techniques, monitoring and early warning systems;
- development of a prototype web-based 'toolbox' of innovative and technically appropriate prevention and mitigation measures;
- case histories and 'hotspots' of European Landslides have been collected and documented;
- research on stakeholder workshops and participatory processes to involve the population exposed to landslide risk in the decision-making process for choosing the most appropriate risk mitigation measure(s).

Project context and objectives:

The growing hazard and risk, the need to protect people and property, the expected climate change and the reality for society in Europe to live with hazard and risk and the need to manage risk are the reasons for initiating the SAFELAND research. The SAFELAND project had three main objectives:

(1) to provide policy-makers, public administrators, researchers, scientists, educators and other stakeholders with improved harmonised framework and methodology for the assessment and quantification of landslide risk in Europe's regions;
(2) evaluate the changes in risk pattern caused by climate change, human activity and policy changes;
(3) provide guidelines for choosing the most appropriate risk management strategies, including risk mitigation and prevention measures.

To achieve these objectives, the research in SAFELAND focused on:

(1) improving the understanding of landslide triggers and run-out and the ability to estimate landslide hazards and risks;
(2) developing a framework for quantitative risk assessment for different landslide mechanisms and different scales and intensity of sliding;
(3) developing risk management tools and guidelines for choosing the appropriate risk mitigation strategy by involving the stakeholders;

Project results:

Work area 1: Landslide triggers and run-out

WP 1.1: Identification of mechanisms and triggers

The activity was aimed at the development of a new characterisation framework for landslides based on triggering processes rather than on landslide material and kinematics. A scheme based on triggering mechanisms helps in identifying the influence of climate change on landslide activity.

Two main activities were carried out:

1) Production of deliverable D1.1. The document is focused on the identification of triggering mechanisms and related landslide types, also as a function of the geo-environmental setting. The work included:

- an extensive and updated literature survey has been used as a starting point to structure the deliverable;
- collection of selected literature and original case studies in order to better explain the state of the art of knowledge and modelling of the landslide triggering processes possibly active in the European Union (EU) context;
- modelling / parametric analysis of specific landslide triggering mechanisms and case histories;
- definition of the geotechnical aspects of landslide triggering;
- analyses of the landslide triggering phenomena due to changes in slope geometry;
- study of the influence of human activities on landslide triggering.

2) Modelling of triggering processes for large rockslides and shallow landslides in soil materials.

Research included:

- for rockslides and debris slides, characterisation and analysis activities aimed at modelling the relationships among hydrological triggers;
- for shallow landslides, the effect of antecedent rainfall, rainfall intensity and rainfall duration on the triggering of landslides have been investigated with physically-based models.

WP 1.2: Geomechanical analysis of weather-induced triggering processes

The aim of WP1.2 was the reporting of an advanced understanding of weather-slope behaviour relationships (objective 1) to be applied for reliable landslide predictions over both spatial and temporal scales (objective 2).

Deliverable D1.2 'Geomechanical modelling of slope deformation and failure processes driven by climatic factors: Shallow landslides, deep landslides and debris flows' is a wide overview on the state of the research around the complex interaction between weather and slopes focusing on the role of weather, which can strongly influence the behaviour of slopes, sometimes leading to catastrophic events.

Deliverable D1.3 'Analysis of the results of laboratory experiments and of monitoring in test sites for assessment of the slope response to precipitation and validation of prediction models' concerns laboratory and field experiments and monitoring sites and aims at providing groundwork for all other deliverables within the work package. D1.3 provides a valuable source of material, organised by both method and soil type, in order to provide a backbone to the work for others to draw from, and for lay readers to learn about known and unknown factors within landslide prediction.

Deliverable D1.4 addresses first as a synthesis of Deliverable D1.2 the different key physical mechanisms in landslides and the mathematical frameworks that have been developed in the soil mechanics community to model the time-dependent processes of water flow, pore pressure dissipation and deformation in soil slopes. Guidelines are presented for landslide modelling which provide the user in the first place with some information on the utility of numerical codes. In the second place, these guidelines allow the user to identify the necessary code components for a given landslide problem, select the necessary and most important field data for the model and perform the modelling steps, including data pre-processing, the actual numerical calculation, as well as the post-processing of the results.

WP 1.3: Statistical studies of thresholds for precipitation-induced landslides

The partners participating in this WP applied existing models for estimating empirical thresholds for precipitation-induced landslides. The types of models used were: intensity-duration, antecedent precipitation, intensity-antecedent precipitation-duration (I-A-D), FLaIR, neural networks and empirical dynamic models. The models were applied to five datasets from France, Switzerland, Italy and Norway. The WP1.3 activities result to deliverable D1.5 which is uploaded on the SAFELAND extranet in its final form.

WP 1.4: Landslides triggered by anthropogenic factors

The main objective of WP1.4 was to improve our knowledge about the impact of human activities on increasing or decreasing the landslide hazard. Landslides can be triggered by both natural and human-induced changes in the environment. Human-induced landslides may result from changes in slope caused by terracing for agriculture, cut-and-fill construction for highways, construction activity, mining operations, rapid draw-down of dams, changes in land cover such as deforestation, and changes in irrigation or surface runoff.

WP 1.5: Run-out models

This WP aimed at presenting models to describe the physical processes involved in landslide propagation, especially in the case of rapid landslides. The results will be used to do complete hazard and risk zonation. The tools allow predicting the velocity and thickness of landslides, as well as the running up along valley flanks and obstacles of different shape and position. The activities contained in this WP 1.5 have been the following:

- joint R&D work aiming at fulfilling the objectives of this WP, which has included visits and stages of personnel belonging to different SAFELAND team;
- there have been contacts between partners in order to write and correct deliverable D1.7;
- the main activity has consisted on collecting information on run out modelling alternatives, to check available codes, and to materialise this information in the deliverable D 1.7.

The deliverable has been structured into the main following sections:

- Mathematical modelling framework, where we have described a set of hierarchically structured mathematical models describing the basic phenomena taking part in propagation phenomena.
- Analysis of different rheological models describing the behaviour of fluidised soils. The analysis was based on studying general 3D rheological models from which depth integrated models could be developed.
- Numerical models for propagation.

One main result of the research and development (R&D) work done in this WP which has been used in WP 1.6 is an analysis of the relative advantages and disadvantages of depth integrated models of Eulerian and Lagrangian types. While both approaches are of similar quality in most of the cases, they present relative advantages and disadvantages which we will describe here. We include some of the conclusions:

- The main advantage of meshless methods, when compared to Eulerian finite elements or finite volume, is the computational cost. In fact, the time of computation is lower than that of classical, Eulerian finite elements, because the computational grid is separated from the structured terrain mesh used to describe terrain topography. The authors of this report have measured differences in computer time of ratios close to 1:30 in favour of SPH when analysing the propagation of lahars in the Popocatepl volcano. The reason is that only a very small part of the topography was occupied by the propagating lahar. In the case of finite elements, all nodes had to be active, hence the much larger cost. In other occasions, as for instance, when studying the propagation of a mudflow originated by the failure of a tailings dam, times are more similar (ratios close to 1:5) Here the reason is that most of the computational domain was occupied by the flow.
- Another aspect which favours SPH methods is the mass conservation, which is enforced in a more effective way. Eulerian finite element models of landslide propagation over long distances suffer from a loss of mass which is much larger than that found in SPH methods. The reason is that Eulerian methods used to make zero heights smaller than a threshold value to avoid numerical instabilities.
- On the other hand, simulation of walls containing a fluid is much easily dealt with finite elements than with SPH methods, which do require special techniques due to their boundary deficiency problem.
- Finally, one important limitation of SPH methods arises when using hydrographs to apply boundary conditions related to the incoming flow in a domain. Indeed, the solution in SPH is to inject nodes, but then we need to apply initial conditions on them, and not boundary conditions.


The purpose of this section has been to illustrate the use of both finite elements and SPH models. The section includes benchmarks which will allow the assessment of run out models. We have considered the following groups:

(i) problems with an analytical solution;
(ii) small scale laboratory tests
(iii) real landslide cases for which we have consistent information.

WP 1.6: Identification of models best suited for quantitative risk assessment (QRA)

The main results from WP 1.6 are synthesised in the two deliverables D1.8 and D1.9 which complement D1.2 D1.3 D1.4 and D1.7 from other WPs. D1.8 and D1.9 provide advice to users on how to work with selected codes, and to warn them of the most frequent error and sources of inaccuracies.

Deliverable D1.8 Guidelines: recommended models of landslide triggering processes and run-out to be used in QRA concerns weather-induced and earthquake-induced landslides. The main aspects to consider in the analysis of the slope response are illustrated, and the difficulty to assess some of the factors required to provide a QRA are clearly stressed. Furthermore, suggestions for a quantitative prediction of landslide triggering are provided accounting for the codes already carefully described in deliverables D1.2 D1.4 and D1.7. An entire part of the report is devoted to earthquake-induced landslides which have not been dealt with in other WPs.

Work area 2: QRA

WP 2.1: Harmonisation and development of procedures for quantifying landslide hazard

Deliverables of WP2.1 consist of compilations of existing resources (landslide databases) and methodologies for QRA.

The basic goal of the work package is to harmonise landslide data bases and procedures for hazard and risk assessment. More specifically:

(a) review the existing databases and propose improvements for achieving interoperability and harmonisation;
(b) review current practices in Europe for landslide mapping, regulations and codes;
(c) provide recommendations for QRA at different scales.

To achieve this goal the following activities have been carried out:

1. Compilation of the European landslide databases (deliverable D2.3)
2. Review of procedures for QRA and preparation of the guidelines for landslide susceptibility, hazard and risk zoning. This task has been split in two phases: the review of the existing practices of landslide hazard and risk assessment in Europe and abroad (deliverables D2.1 and D2.2) and preparation of guidelines for QRA (deliverable D2.4).

The objective of deliverable D2.1: 'Overview of current landslide hazard and risk assessment practices in Europe' was to review the current practices, regulations and codes in Europe for landslide mapping, susceptibility, hazard and risk assessment. The contents of this deliverable refer to the existing official practices that are currently promoted or applied by administration offices, geological surveys, and decision makers:

- The classification criteria for landslide types and mechanisms present large diversity even within the same country. As a result in some cases no landslide mechanisms are specified and in some others there is an exhaustive list. Each mechanism requires its own method of assessment. The differentiation of landslide types and mechanisms is recommended particularly for scales larger than 1:25 000.
- In relation with the types and mechanisms of landslides represented in the maps, in general they are grouped in a few general mechanisms (i.e. rockfalls, slides, flows).
- The effect of hazard amplification due to the spatial superposition of different types of instabilities should also be taken into consideration, as well as the synergistic action of other natural phenomena (i.e. earthquake) wherever applicable, regardless of the mapping scale.

Additionally, within the framework of the WP2.1 a workshop was organised in the Chengdu University of Technology on 13 and 14 April 2010, with the aim of assessing the state of art of landslide hazard and risk assessment in the Peoples Republic of China.

Deliverable D2.2 included contributions from China and India summarising landslide risk assessment work from their respective countries.

Deliverable D2.3 'Overview of European landslide databases and recommendations for interoperability and harmonisation of landslide databases' made a detailed review of existing national landslide databases in Europe together with a number of regional databases and proposed improvements for delineating areas at risk in agreement with the EU soil thematic strategy and its associated proposal for a Soil Framework Directive, and for achieving interoperability and harmonisation in agreement with Inspire Directive. The report was based on the analysis of replies to a detailed questionnaire sent out to the competent persons and organisations in each country, and a review of literature, websites and main European legislation on the subject.

Deliverable D2.4: 'Guidelines for landslide susceptibility, hazard and risk assessment and zoning', has been prepared aiming at:

- recommending methodologies for the quantitative assessment and zoning of landslide susceptibility, hazard and risk at different scales (site specific, local, regional and national);
- proposing specific methodologies for different landslide mechanisms;
- including a selection of the best suited procedures for verification of the models and validation of the results.

WP 2.2: Vulnerability to landslides

Research activities have resulted in the finalisation of all methodologies for assessing the vulnerability of different elements at risk (buildings, lifelines, persons etc.) exposed to different landslide hazards. Both physical and socioeconomic vulnerability is addressed. Some of the developed methodologies were also applied to selected real case studies. The outcome of WP 2.2 activities is the three deliverables D2.5 D2.6 and D2.7.

The most significant results in WP2.2 within the reporting period are summarised as follows:

- Proposition of a generic vulnerability model which can be used for a large portfolio of rockfall protection galleries. The methodology includes three main steps:

(a) definition of the exposure for rockfall protection galleries;
(b) resistance modelling for rockfall protection galleries;
(c) development of vulnerability curves for rockfall protection galleries; each of these involves various sub-steps.

Different sources of uncertainty can be included in the analysis in a quantitative cost-effective manner (ETHZ).

- Literature review on existing models for the quantification of physical vulnerability of persons exposed to different landslide hazards. The most important factors concerning the different aspects of physical vulnerability of persons to landslides are discussed (AUTH).
- Comparison of the derived curves for roads to debris flows developed within the 1st reporting period by TRL-AUTH-UPC with real debris flow events from both Scotland in the United Kingdom (UK) and the Republic of Korea (TRL).
- Assessment of the physical vulnerability in terms of building's (homogeneous) aggregates due to different slow moving landslide hazards at the territory of the National Basin Authority of 'Liri-Garigliano' and 'Volturno' rivers, central-southern Italy at small scale (1:100 000). Moreover, the vulnerability of smaller building aggregated levels affected by slow movements at two study areas (scale 1:25 000) within the already investigated territory is assessed (UNISA).
- Estimation of the vulnerability of buildings subjected to rainfall induced slow moving landslides located at the test site of San Pietro in Guarano, Cosenza Province, southern Italy (scale 1:2000) based on the corresponding methodology developed in D2.5 (UNISA).
- Investigation of the physical vulnerability of a representative RC building subjected to earthquake triggered slow moving landslide hazards located near the Kato Achaia (western Greece) slope's crest to assess the validity of the derived fragility curves proposed in D2.5 (AUTH).
- Assessment of the vulnerability of the roadway system of Grevena in Greece due to earthquake triggered landslides based on the proposed fragility functions developed in D2.5 (AUTH).
- Improvement of the socio-economic vulnerability model by introducing three new indicators referring to critical infrastructure, risk awareness and early warning capacity (ICG).
- The socio-economic vulnerability model is applied to four more locations (expect for the two in Norway already implemented during the first reporting period): Andorra-Spain / France, Barcelonnette -France, Grevena-Greece, Romania. (ICG).

WP 2.3: Development of procedures for QRA at regional scale and European scale

Three deliverables, D1.8 D1.9 and D2.11 have been produced in WP 2.3. Deliverable D2.8 'Recommended procedures for validating landslide hazard and risk models and maps' proposes methods for the following:

- Quantification of the reliability of the assessment, accounting for:

i. data vagueness and uncertainties (relevant to landslide inventory and conditioning factors and material constitutive parameters);
ii. accounting for the 'limited' knowledge on the physics of the processes (relevant to hydro- and mechanical understanding of the mechanisms and process modelling);
iii. taking into account the issue of the 'mapping unit', independently of the scale.

- Quantification of the validity of the assessment, considering:

i. validation / evaluation of the maps;
ii. the multi-criteria problem of adequacy (conceptual, mathematical) in describing the system behaviour, robustness to small changes of the input data (i.e. data sensitivity), and accuracy in predicting the observed data;
iii. the type of the output (susceptibility / hazard / risk).

The proposed methods are summarised to:

- Methods and measures to quantify the reliability of the assessment.
- Methods and measures to quantify the validity of the assessment.

Deliverable D2.9: 'A toolbox for landslide QRA' is composed by three tools (computer applications) and a manuscript that is addressed to the end-users and it includes the description of each toolbox, the prerequisite inputs, the obtained outputs, the followed methodology and possible limitations for its use. The three tools serve at:

- landslide QRA
- rockfall quantitative vulnerability of buildings
- rockfall QRA.

The objective of Deliverable D2.11 'QRA case studies at selected 'hotspots' was to present some practical applications of QRA that may serve as examples that might be followed by scientists and practitioners depending on the afore-mentioned factors (landslide type, scale, risk descriptors etc). The added value of them is that, in comparison with the current state-of-the art, they incorporate innovations related to the calculation and hazard and vulnerability in order to incorporate them into the risk assessment. The goal of this deliverable is to cover a range of different cases as far as it concerns:

- the application scale: regional, local, site-specific;
- the landslide type: debris flow, deep-seated landslides, hyper-concentrated flows, rockfalls;
- the source of input data: empirical to remote sensing;
- the inclusion or not of the run-out modelling;
- the vulnerability assessment: buildings or people, empirical or analytical, deterministic or probabilistic, element at risk-oriented (detailed) or generalised;
- the used risk descriptors: qualitative or quantitative, and in what terms.

To this end, five-case studies were presented in this deliverable:

- debris slides - rapid earthflows at Castellamare de Stabia, Naples province, Italy;
- deep-seated landslide in Ancona, Italy;
- hyperconcentrated flow at Nocere Inferiore, Italy;
- rockfalls at Andorra;
- rockfalls at Fiumelatte, Italy.

WP 2.4: Identification of landslide hazard and risk 'hotspots' in Europe

The objective of WP2.4 is to perform a first-pass analysis of landslide hazard at European scale to identify the landslide hazard and risk 'hotspots', where hazard and risk are highest.

Hotspots of landslide hazard and risk were identified by an objective geographic information system (GIS) based analysis for Europe. The results show clearly where landslide pose the largest hazard in Europe and the objective approach allows a ranking of the countries by exposed area and population. In absolute numbers Italy is the country with the highest amount of area and population exposed.

The results can be found at the final version of the deliverable D2.10: 'Identification of landslide hazard and risk 'hotspots' in Europe'. The final version also includes a detailed description of the applied models and a discussion of the differences between model results.

Work area 3: Global change scenarios

WP 3.1: Climate change scenarios for selected regions in Europe

Different regional climate model simulations over Europe (from the EU Sixth Framework Programme (FP6) project ENSEMBLES) at a spatial resolution of 25 x 25 km2 have been used to perform an extreme value analysis for trends in heavy precipitation events. Furthermore, potential causes for trends in heavy precipitation have been investigated by analysing a variety of thermodynamic and dynamic variables as simulated by the regional climate model REMO. Deliverable D3.1 reports on the performed work.

Climate change simulations with the regional climate model REMO are performed for three selected regions over Europe: Italy and the Alps, Northern Europe, and Eastern Europe. The simulations have been carried out for the time period 1951 - 2050, employing the SRES A1B emission scenario.

These climate simulations are delivered to the Centro Euro-Mediterraneo per i Cambiamenti Climatici (CMCC) in order to be used as boundary conditions for further model simulations. The usage of the model output data for simulations on an even more refined grid is expected to improve the ability to simulate even localised heavy precipitation events in regions where rain-induced landslides occur on a regular basis.

The climate simulations are presented in Deliverable D3.3. They have been performed by CMCC on four selected areas in Europe, using the regional model COSMO-CLM. The regions are:

1. Nedre Romerike, Southern Norway;
2. Pizzo d'Alvano, Campania, Italy;
3. Barcelonnette, French Alps;
4 Telega, Romania.

Results of two-meter temperature and total precipitation averaged over the time periods 1971 - 2000 and 2021 - 2050 have been presented, in order to highlight the variations expected in the future, with respect to the past period. Furthermore, an extreme value analysis for projected future changes in heavy precipitation is carried out for the different regions and separately for summer and winter.

Finally, deliverable D3.4 presents a synthesis and discussion of the results. The analysis concentrates on projected future changes in heavy precipitation for four target regions in Europe: Southern Norway, southern Italy, the Alps and Romania.

Main results D3.1:

- In winter we see a general trend towards more heavy precipitation events across all analysed regional climate model simulations. This could partly be explained by an increased amount of vertically integrated water vapour.
- For summer, we could find a slight increase of heavy precipitation in Northern Europe and a general decrease in southern Europe in all regional climate model simulations. The models suggest an increase of the air temperatures all over Europe and particularly in the southern part. Nevertheless, the trend of the vertically integrated water vapour does not follow the temperature trend linearly.

Main results D3.2

- The strongest warming is found in the southern regions in summer and over cold regions in spring and autumn, where the warming is amplified due to the snow-albedo feedback. Precipitation is projected to increase in cool and moderate regions, but decreases in the warm regions during the warm seasons.

Main results D3.3

- In the area of Nedre Romerike (Norway) strong increases of temperature are projected especially in winter, while a general increase of precipitation is expected in winter, with a general increase of extreme events which is most pronounced in the western part of the domain.
- In the area of Pizzo d'Alvano (Italy), a growth of temperature is also projected, even if less evident than the previous case. In winter, strong increases of precipitation (with strong extreme events) are expected in the area of Pizzo d'Alvano, In summer slight reductions are expected for the average monthly precipitation over the whole domain, which is in contrast to a projected increase in daily precipitation extremes in the Pizzo d'Alvano region and along the western coast line.
- In the area of Barcelonnette (France) significant increases of temperature are expected in the future, up to 3 degrees Celsius, in both seasons, but especially in winter. An increase of precipitations is expected in small sub domains in both seasons, with slight changes of extreme events on the whole domain.
- In the area of Telega (Romania), a general increase of temperature of about 1.5 degrees Celsius is expected over the whole domain, for both summer and winter. In winter an increase of precipitation is expected, while a general significant reduction is expected in summer; an increase of extreme events is expected in winter and summer in the north of the domain with the magnitude of the changes being higher in winter.

Main results D3.4

- Both the analyses from D3.1 and D3.3 show mainly positive trends of heavy precipitation in winter. Strong changes are particularly found in mountainous regions, where the impact on landslides may be large. The summer trends in Northern Europe are generally weaker than the winter trends. In warm and rather dry regions, such as the Pizzo d'Alvano domain, which is located in Campania in South-Western Italy, or the Telega domain in Romania, the average summer precipitation is projected to decrease. In other words, dry regions tend to become even drier. For extreme events on the other hand, increasing trends are found for some of these regions, in particular in the high-resolution COSMO-CLM simulation. This indicates that especially the typical convective events in summer may occur with higher probability. For regions where the average precipitation decreases and the soil dries, the drainage of the soil is reduced and consequently the occurrence of heavy precipitation events may have strong impacts, such as flooding. On the other hand, regions which already possess a moist climate, such as Southern Norway, tend to become even wetter on average and also in the extremes.

The results of WP 3.1 are used in WP 3.3 in particular in deliverable 3.7.

WP 3.2: Human activity and demography scenarios

The initial objectives of WP 3.2 were to provide information on prospective human activity and demography evolution in Europe and in selected sites in correspondence with IPCC scenarios at four selected dates: 2030, 2050, 2070 and 2100. In order to provide such data, gathering, compilation and interpretation of available data had been necessary at both national and European level.

Task 1: European and national levels

The objective of deliverable D3.5 was to respond to the following core questions:

1. What are human activities and population characteristics that affect or are affected by landslides?
2. What recent and existing scenario projects are relevant for the activities of the SAFELAND project?
3. What relevant data are available for which scale, and are there forecasts for the years 2030, 2050, 2070 and 2100?

Based on results from other work packages, anthropogenic factors modify the three dimensions of landslide risk: exposure, hazard and vulnerability and the main factors that have an influence are demography, economics, and land use / land cover. To fulfil the second and third objectives, global scenarios (project context, scenario specification, results and limitations), European scenarios (project context, scenario specification, results and limitations) and some national scenarios have been reviewed. For each scenario reviewed, the context, the scenario specifications and the results and limitations have been presented.

The main result of this review was to stress that an abundant supply of data on all kinds of issues is available throughout Europe and the rest of the world. However, most countries and organisations that collect the data use different definitions of indicators, different methodologies and different territorial units. Depending on the region, data is more or less complete and reliable.

Task 2: Local sites

Subsequently, the evolution of human activity factors impacting landslide risk from 2030 till 2100 and at the level of test sites has been considered. The idea was to check data availability at the level of selected hotspots. When they exist, prospective data are used. Unfortunately, data are sparse, rarely spatialised and not always adapted to the local context. However, this lack of information can be partially compensated by the analysis of past and present trends. Satisfactory data have been collected for the Barcelonnette site and have allowed the elaboration of demography scenarios at local level by 2030. The land use change scenario by 2100 has been studied.

Acknowledging significant uncertainties, the demographic forecasts can be extended from 2030 to 2100. The economic changes scenario has not been treated as such a scenario is really difficult to implement at a local scale and also to integrate in risk analysis process. Demographic scenarios have been partially developed for the Nedre Romerike site (Norway).

WP 3.3: Landslide risk evolution in selected 'hotspots' areas

The initial objectives of WP 3.3 were to:

1. provide updated risk maps according to global change at four 'hotspots', representative of different landslide types and contexts in years 2030, 2050, 2070 and 2100;
2. analyse of impacts of risk evolution due to global change for mitigation strategies and risk management purposed on 'hotspots'.

Integration of the climate change scenarios:

This first objective has been mainly addressed in deliverable 3.7: 'Expected changes in climate-driven landslide activity (magnitude, frequency) in Europe in the next 100 years and deliverable 3.8 - Changing pattern in climate-driven landslide hazard at selected sites Europe (focus on southern Italy, the Alps, and Southern Norway) in the next 50 years'. The European-scale analysis of present and future landslide hazard and risk has required many simplifications. The main difficulty was to find homogenous datasets that cover all of Europe with the same accuracy. This problem is even increased when the datasets have to cover future predictions.

The climate model results used in this study are based on a physical climate model and have a reasonable level of uncertainties in the future predictions. On the other hand, land cover and population datasets are secondary products based on climate simulations and economical modelling, which naturally include more errors in the process and are far more uncertain. In this context, the predicted changes in landslide hazard and risk in Europe, although certainly indicative, have to be investigated and used with care.

The climate change scenario of WP 3.1 has been integrated in landslide hazard assessment at site specific scale in deliverable D3.8. The impact of climate change on landslide hazard has been assessed on the three focused areas. The data provided for the Romanian and Spanish sites were not sufficient to realise the full study of hazard assessment. In order to provide estimates for the Spanish site, the lack of climatic data has been dealt with by using climatic data from another regional climate model.

Different methods (statistical, empirical and physically-based methods) have been used on the different sites. Even if these sites present different contexts in view of landslides causes (climates, size of landslides), the analyses show that climate change is likely to induce similar trends in landslide activities. Based on the IPCC A1B scenario and on the resulting climate change scenario at local scale, the different models predict a very increase in landslide activities. This change would materialise either as an increase in the frequencies of landslides or as an increase in surface area of the potentially unstable areas. However, these models require precise data, not only for calibration but also for prediction, and so climate models should be adapted to such resolutions, like in this study.

The results differ from the predictions provided by larger scale models. These differences might be explained by the finer calibration processes used for local scale analysis and also to the finer climate model used, which, for example, take into account the influence of topography on climate (mostly on precipitation). So, if large scale models are useful to determine where landslide activities will vary relatively to the other regions, the different kinds of local scale models are necessary for urban planners and all local authorities to estimate what would be the future risks in their communes or valley, with for some of the models, spatial information. However, these models require precise data, not only for calibration but also for prediction, and so climate models should be adapted to such resolutions, like in this study.

Integration of land cover and human activity exposure scenarios and related vulnerability on selected 'hotspots'

This objective has mainly been targeted by deliverable D3.9 - 'methodology for predicting the changes in the landslide risk during the next 50 years at selected sites in Europe. Changing pattern of landslide risk in hotspot and evolution trends Europe according to global change scenarios.'

The potential effect of climate change on landslide triggering depends on the type of landslides considered. For rainfall induced landslides, the hazard evolution is tightly linked to the variation of precipitation in time as threshold parameters. Other types of landslide may be impacted differently by climate change, for instance rock fall main triggering parameters are the frost and defrost cycles; but those cases were not developed in the work performed within Work area 3. This work presented means of assessing landslide risk evolution with climate change scenarios. Those methods depend on the availability of input data. If climate change scenarios and land cover evolution scenarios can be developed quite accurately; the scenarios of population and human activity evolution are rougher; especially at site scales.

Nevertheless three studies of landslide risk assessment have been performed on French, Norwegian and Scottish sites. The results seem to show a similar trend: an increase of landslide risk which is more or less significant depending on the considered sites. Due to a high level of uncertainties on population and traffic evolution scenarios, precautions need to be taken when interpreting and using the results.

Some new avenues of research for a more precise assessment of the future landslide risk evolution at site specific scale have been investigated. A methodology has been developed to obtain simplified and rapid estimations of the influence of suction change due to vegetation/canopy on the factor of safety. The approach considers only the influence of the root water uptake caused by evapotranspiration and other phenomenon due to the presence of vegetation is not taken into account. In parallel, time-dependent fragility analysis of corroded RC buildings impacted by co-seismic permanent landslide displacements has been developed. In the future, this methodology could to be adapted to rainfall triggered landslides in order to evaluate risk evolution as a combination of hazard changes, exposure changes and vulnerability changes.

Work area 4: Monitoring technology

WP 4.1: Short-term weather forecasting for shallow landslide prediction

A model has been developed being a combination of an infinite slope stability calculation with a transient, analytic solution for pore pressure response to steady state and transient rainfall infiltration. A complex operative chain has been set up based on forecasted rainfall, by means of the COSMO-Lokal-Modell (LM) model, a hydrological model and a simple infinite-slope stability code. The code has been developed coupling an infinite-slope stability analysis with a three-dimensional (3D) analytical solution for transient pore pressure response to steady state and transient rainfall infiltration, allowing a regional slope stability evaluation in a GIS framework. The numerical weather prediction model used to produce forecast rainfall is COSMO-LM. The LM is a non-hydrostatic limited-area atmospheric prediction model.

The definition of a prototype tool for early warning of rainfall-induced landslide has been completed. A large number of test cases have been simulated, providing an adequate verification of the tool together with knowledge about its weak and strong points. At the same time research activities for the development of the different simulation codes and of the software linking the different simulation models has been continued by AMRA, CMCC and UNIFI. The implemented tool, that simulates the 'hydrometeorological simulation chain', has the goal to evaluate the modification in the slope safety factor (FS) using precipitation forecast coming from numerical weather prediction models. The tools consists of several components: numerical weather prediction (NWP) models, tools for precipitation downscaling techniques realising the link between atmospherical models and stability analysis models and software codes for the production of safety maps on scales ranging from individual slopes to regional level.

The tool has been tested on the instrumented Cervinara site, the Tuscan region and on the Ischia isle. The choices of the test cases is mainly due to the availability of well documented data in the selected area about soil structure, properties, and initial conditions together with hourly precipitation observations from different in situ stations. The test cases covers very different meteorological situations such as advection rain events, with a long time period, convective rainfall happening in a short time periods and also days in which the rainfall is not intense. With regard to the stability models, the test cases covered various spatial scales and soil with different properties. Generally, quite good agreement was found between the forecast and the observation of the soil conditions after precipitation. Some problems need to be solved such as the effect on the soil of snow melting (increasing of the water availability at the soil level).

WP 4.2: Remote sensing technologies for landslide detection, monitoring and rapid mapping

The WP4.2 has the general objective of analysing the use of remote sensing imagery (Spaceborne radars, airborne and VHR space borne optical sensors and airborne geophysics) in landslide studies. In particular the specific objectives are:

- define and validate a common methodology for detection, rapid mapping, characterisation and monitoring of landslides at regional and catchment scales using advanced remote sensing techniques;
- define and validate a common methodology for the rapid creation and updating of landslide inventories and hazard maps at regional / catchment scale using advanced remote sensing techniques;
- prepare user-oriented guidelines for the incorporation of advanced within integrated risk management processes and best practices.

The WP deliverables are the following:

• D4.1 - Review of monitoring and remote sensing methodologies for landslide detection, fast characterisation, rapid mapping and long-term monitoring;
• D4.3 - Creation and updating of landslide inventory maps, landslide deformation maps and hazard maps as input for QRA using remote sensing technology;
• D4.4 - Guidelines for the selection of appropriate remote sensing technologies for monitoring different types of landslides;
• D4.5 - Evaluation report on innovative monitoring and remote sensing methods and future technology (together with WP4.3).

D4.1 - Review of monitoring and remote sensing methodologies for landslide detection, fast characterisation, rapid mapping and long-term monitoring

This review aims at representing a common reference for the different deliverables of SAFELAND work area 4. In addition to being a state-of-the art overview, this deliverable provides helpful and extensive support for non-specialists and students interested in the application of new techniques to different mass movements. The core of this deliverable consists of two main chapters (2 and 3), which aim at developing the basic technical knowledge for:

(a) landslide detection,
(b) fast characterisation,
(c) rapid mapping and
(d) long-term monitoring.

D4.3 - Creation and updating of landslide inventory maps, landslide deformation maps and hazard maps as input for qra using remote sensing technology

The deliverable provides a comprehensive view on the latest developments of remote-sensing technologies as applied for the creation and updating of landslide inventory and deformation maps by the members of the SAFELAND WP 4.2. Furthermore, chapter 4 gives a broad overview of input datasets for hazard and risk assessment that can be obtained through remote sensing, and in chapter 5 suitable updating strategies as well as steps toward a better linkage between the recent technological developments and QRA methods are discussed.

D4.4 - Guidelines for the selection of appropriate remote sensing technologies for monitoring different types of landslides

This document provides condensed guidelines for the selection of the most suitable remote sensing technologies according to different landslide types, displacement velocities, observational scales and risk management strategies. The main part of the document gives an overview of the capabilities of different techniques to detect, characterise, map and monitor landslides and can be used to initially constrain the choice of methods to a few techniques that seem most feasible for the landslide process at hand. Before final decisions on the methods to be used are taken, further information and expertise will typically be required.

Users of this document should consider that it provides a snapshot of the currently available knowledge and technology. In the near-future, the launch of new satellites, better data access (e.g. global monitoring for environment and security (GMES)), lower data prices and on-going enhancement of processing algorithms, will lead to the maturing of many currently new or experimental techniques into methods suitable for operational use.

D4.5 - Evaluation report on innovative monitoring and remote sensing methods and future technology

This deliverable is a joint deliverable between WP4.2 and WP4.3 and had the aim of making an evaluation of the most innovative landslide monitoring and remote sensing technologies used at present, as well as suggesting needs for research and technical developments of the existing methodologies. Amongst all the ground based techniques employed in landslide studies, the ones which in recent years showed the most promising improvements were selected and reviewed, emphasising the recent trends in their development and application and stressing the latest scientific and technological advances. The same approach was pursued with remote sensing techniques, making a clear distinction between the use for detection and mapping and the use for monitoring purposes.

The objectives of the deliverable were achieved through these main steps:

- Overview of recent and emerging ground based techniques for landslide analysis.
- Overview of recent and emerging remote sensing technologies for landslide analysis.
- Questionnaire on landslide monitoring methods. The questionnaire on national state of landslide site investigation and monitoring was prepared and was disseminated among European institutes and representatives within the frame of the SAFELAND project. The results of the questionnaire were reported and discussed.
- Questionnaire on remote sensing technologies. The aim of the questionnaire was to collect information about the usefulness of remote sensing for landslide study and to evaluate its applicability for landslide detection, mapping, monitoring and early warning. This questionnaire was circulated within and outside SAFELAND consortium. The results of the questionnaire were reported and discussed.
- A relevant part of the deliverable was focused on the application of these innovative techniques within SAFELAND case studies, clearly stating which technical and scientific improvements were achieved for each technique thanks to SAFELAND project.
- Evaluation of ground based, airborne and space-borne techniques based on the literature review, on the aforementioned questionnaires and on the results coming from the SAFELAND case studies.

WP 4.3: Evaluation and development of reliable procedures and technologies for early warning

- Task 1: Assessment of current state-of-art in monitoring and early warning (technology)

The current state-of-art in monitoring and technology of early warning has been proceeded through:

• contributing to deliverable D4.5 (Remote sensing technologies for landslide detection, monitoring and rapid mapping/ Evaluation and development of reliable procedures and technologies for early warning, Responsible: UNIFI/ITC, Delivery month: 24).
- A questionnaire study on 'National mass-movement investigation and monitoring', and 'Questionnaire on remote sensing' focused especially on the use and reliability of field investigation, remote sensing, and monitoring techniques for landslides which were presented in deliverables D4.5 and D4.6 ('Report on evaluation of mass movement indicators', Delivery month: 32).
- A screening study of existing EWS systems worldwide, which has been made for deliverable D4.8 'Guidelines for monitoring and early warning systems in Europe (Delivery month: 32).

- Task 2: Exploring the role of 'geo-indicators' (mass movement parameters) as early warning parameters (processes and related parameters)

The task on role of monitoring and early warning parameters is the main goal of the deliverable D4.6 'Report on evaluation of mass movement indicators'. This task was mostly based on analysis and evaluation of monitoring field data of unstable slopes at SAFELAND test sites. The project partners have been collecting the raw monitoring data and provided their analysis from 14 test sites through Europe.

- Task 3: Method evaluation and implementation of guidelines for monitoring and early warning

This task is mostly based on the practical implementation of early warning systems. It is presented in D4.7 and D4.8 and it summarises the theoretical and practical information to derive general rules to provide effective EW centres worldwide.

The outcome of the work has been reported in four deliverables:

D4.5: Evaluation report on innovative monitoring and remote sensing methods and future technology (together with WP 4.2)

D4.6: Report on evaluation of mass movement indicators
The deliverable 'Report on evaluation of mass movement indicators' focused on physical parameters which could be monitored in relation to landslide triggering processes, and which could potentially be used as early warning parameters of slope instabilities. The first part of the report reviewed potentially available monitoring parameters, including basic definitions, units, typical values, and formulas. Not only the well-established parameters were included, but also those with a potential for a future application as early warning parameters. Based on a questionnaire study, the parameters were evaluated by means of their abundance, reliability and early-warning potential. The third part of the deliverable presented and summarised results from the SAFELAND test sites. The monitoring results from the WP 4.3 provide an excellent basis for future research in the field of early warning parameters and thresholds. However the results of the deliverable show that there is a need for long-term monitoring experiments, exceeding the three year period of SAFELAND.

D4.7: Report on the development of software for early-warning based on real-time data

This deliverable described new software specifically developed to support technical staff in data analysis and the decisional process. The project intended to realise a centralised interface for early warning centres to manage data from different monitoring stations. The proposed software should be a separate and independent tool for real-time analysis and evaluation of geo-scientific monitoring data, including threshold evaluation. The basic concept was to develop software that could integrate and automatically analyse data from a variety of sensors. The integration and predefined analysis and correlation of different sensors would help the user in operative early warning centres to increase the quality of the geo-scientific evaluation. As a result, this report gives a brief description of the application structure and all necessary steps to start up a system. Finally, this report describes the data analysis in one of the test sites included in this part of the project.

D4.8: Guidelines for monitoring and early warning systems in Europe - Design and required technology
The D4.8 deliverable summarised how landslide early-warning systems should be designed and operated and presented a screening study of existing EWS systems worldwide. The document was elaborated as the last deliverable of work area 4 and aimed at facilitating the decision process for stakeholders by providing guidelines. For the purpose of sharing the globally accumulated expertise, a screening study was realised amongst 14 early warning systems. As a result, the report presented a synoptic view of existing monitoring methodologies and early-warning strategies and their applicability for different landslide types, scales and risk management steps. Several comprehensive checklists and toolboxes were also included to support informed decisions.

Work area 5: Risk management

WP 5.1: Toolbox for landslide hazard and risk mitigation and prevention measures

Deliverable D5.1: Compendium of tested and innovative structural, non-structural and risk-transfer mitigation measures for different landslide types
A categorisation system for the different structural mitigation and prevention measures was developed and a total of about 60 measures were selected for further documentation and evaluation. The draft report on the website describes a number of these measures with a brief discussion of the classification of the possible mitigation measures; guidance on the applicability and effectiveness of each mitigation measure considered to different types of landslides; information on the maturity of the technology, which can range from 'prototype development' to 'obsolete'; information on current design methods, their maturity and associated uncertainties; and comparative (qualitative) information on costs. The measures are evaluated for different types of ground movements and slides. Each measure was then ranked with 'scores' and placed into an applicability matrix that will be used later in the toolbox. The report on the compendium is on the SAFELAND website.

Delivery D5.2: Web-based toolbox of structural and non-structural mitigation measures with decision-making guidance
A piece of web-based software was prepared and is now being tested. The toolbox is implemented for local landslide hazards and includes typical examples. It assists the user with the following:

- selecting the type of ground movement expected;
- assessing the level of hazard associated with the ground movements;
- evaluating the consequences of the ground movement;
- evaluating the risk class and determining the need for mitigation; and
- for selecting the most appropriate mitigation approaches to use, and comparing them.

The toolbox contains default implementation criteria and 'scores' for each mitigation and preventing measure. The user can at any time introduce his own 'scores' and 'weights' for each of the mitigation measures.

Delivery D5.3: Quantitative risk-cost-benefit analysis of selected mitigation options for two case studies

Decision making in general is a difficult issue due to the significant underlying uncertainties and complex interrelation of events and choices affecting the benefits and losses associated with decisions. Typical decision problems are subject to a combination of inherent, modelling and statistical uncertainties. This is primarily due to the fact that the understanding of the issues involved in the decision is often incomplete and that the processes of physical phenomena and human interactions can be modelled only in uncertain terms. If all aspects of a decision problem would be known with certainty, the identification of optimal decisions would be straightforward by means of traditional cost-benefit analysis. Due to the existing uncertainties, it is not possible to assess the results of decisions in certain terms. There is hence no way to assess with certainty the consequences resulting from the decisions we make. However, what can be assessed is the risk associated with the different decision alternatives. Based on risk assessments, decision alternatives may then be consistently ranked on the basis of their associated utilities and benefits/losses, thereby providing a rational basis for societal decision making. This report provides a framework and methodology for carrying out a risk-cost-benefit analysis for decision-making. Two case studies applying the proposed methodology - one involving the analysis and management of risks arising from debris flow phenomenon in Barcelonnette, and the other with risk analysis and risk management for risks posed by different flow-like phenomena in Nocera Inferiore are described in the report.

The report concludes that risk assessment and risk management can be seen as an essential and integral part of the decision planning, decision support and decision-making processes. Decision problems in general and especially in natural hazards management are generally subject to a combination of inherent, modelling and statistical uncertainties. What can be assessed is the risk associated with the different decision alternatives. Based on risk assessments, decision alternatives may then be consistently ranked on the basis of their associated utilities (which may be more useful for engineering decision problems) and cost-benefit analyses (which may be relevant for life safety and overall risk management problems), thereby providing a rational basis for societal decision making. The proposed framework for carrying out a risk-cost-benefit analysis for decision making provided convincing results for the two case studies. The usefulness of the life quality index (LQI) approach for the evaluation of the acceptance of the mitigation options with regard to investments into life safety and the evaluation of the optimal risk mitigation alternative was demonstrated.

Delivery D5.4: Quantification of uncertainties in the risk assessment and management process

The consideration and treatment of uncertainties is an essential part of any risk assessment and risk management process. Uncertainties can either be naturally inherent or modelling and statistical related. This deliverable provided a rational basis for the quantification of the different uncertainties existent in the risk assessment and risk management processes. To obtain a complete picture of the issues and aspects concerning the treatment, quantification and management of uncertainties in the risk assessment, risk management and decision making processes, this deliverable report should be read in conjunction with the report of deliverable D0.3. The deliverable presented an example on the modelling and management of uncertainties associated with rock fall hazards following a Bayesian approach.

The report concluded that a rational and consistent understanding and consideration of uncertainties is vital for any risk assessment and risk management process and for ensuring rational and optimal decision-making. It is hence necessary to think about the nature of the various types of uncertainties, particularly in the context of risk communication.

WP 5.2: Stakeholder process for choosing an appropriate set of mitigation and prevention measures

WP 5.2 activities included the following deliverables:

D5.5 Five scoping studies of the policy issues, political culture and stakeholder views in the selected case study sites - Description of methodology and comparative synthesis report

This period of work focused on the continuation of efforts set in place during the previous reporting period where project researchers for each of the five case studies were recruited, methodology determined in a London workshop and synergies across the WP explored in a meeting at IIASA. Continuing work led to the completion of data collection for each of the five case study sites and the synthesis of these reports into a final project document and deliverable which was submitted to schedule.

D5.6 Development and testing of spatial multi-criteria evaluation for selected case sites

This resulted in the first draft of the deliverable document that was eventually submitted in April 2012. Currently the main contributors are reworking the document into a paper that should be published along with other SAFELAND papers.

D5.7 Design and testing: A risk-communication strategy and a participatory process for choosing a set of mitigation and prevention measures

The core research work for this deliverable has been performed during this reporting period. It included:

- the organisation of the participatory process (1 public open meeting, 5 meetings with 15 selected residents, evaluation and feedback via questionnaire, informal meetings with local activities, parallel working groups);
- a questionnaire survey (questionnaire piloting, collection of 373 questionnaires);
- communication and education activities (setting up of a website, online discussion group, press releases and contacts with local media, simulation exercise with students at the LARAM school organised by UNISA).

The results of this work were synthesised in the deliverable which includes a detailed description of the case study, the methodological approach, and the key results of: the qualitative work, the participatory process, the questionnaire survey and the communication and education activities.

WP 6: Demonstration sites and case studies for verification/calibration of models and scenarios

The main objective of WP6 is to document case histories and 'hotspots' of European Landslides (including potentially unstable slopes), and to provide the technical data for the case studies to be used in other work packages in SAFELAND.