Natural Science
Vol.6 No.3(2014), Article ID:43237,14 pages DOI:10.4236/ns.2014.63018

Development of methodology for quantitative landslide risk assessment—Example Göta river valley

Yvonne Andersson-Sköld1,2,3*, Stefan Falemo1,4, Marius Tremblay1,5

1Swedish Geotechnical Institute (SGI), Göteborg, Sweden; *Corresponding Author: yvan@cowi.se

2Department of Earth Sciences, University of Gothenburg, Göteborg, Sweden

3COWI AB, Göteborg, Sweden

4ÅF Infrastructure, Göteborg, Sweden

5Tyréns, Geotechnical Engineering, Göteborg, Sweden

Copyright © 2014 Yvonne Andersson-Sköld et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. In accordance of the Creative Commons Attribution License all Copyrights © 2014 are reserved for SCIRP and the owner of the intellectual property Yvonne Andersson-Sköld et al. All Copyright © 2014 are guarded by law and by SCIRP as a guardian.

Received 31 December 2013; revised 31 January 2014; accepted 7 February 2014

KEYWORDS

Landslide; Risk Assessment; Methodology; Climate Change

ABSTRACT

Effective landslide risk management requires knowledge of the landslide risks. This paper presents a risk assessment methodology for semiregional scale. The landslide probability is assessed taking into account expected climatechange in the case study area (the Göta river valley). Climate change is expected to result in increased erosion and water fluctuations. There are large areas with marine clays, often quick clay, in the area and the landslide process can be rapid with extensive damages and casualties. The consequence methodology includes a wide range of consequences assessed by monetary valuation. The consequences and the landslide probability are combined as pairs of values in a risk matrix and the risk is also presented on a map. The map has been used as discussion and decision bases in the municipalities in the Göta river valley, in the county administration and on governmental level to estimate the needs of risk mitigation and to make priorities.

1. INTRODUCTION

Climate change is expected to cause severe impacts on the natural and built environment. In addition to the temperature increase per se, climate change is expected to profoundly influence the hydrology, e.g. [1]. In Norway, Sweden and Northern Europe, the annual precipitation is expected to increase by up to 30% in the period 2071- 2100 compared to the reference period 1961-1990 [2-4]. The river runoff is in general expected to be higher in winter, followed by a less pronounced snowmelt peak and lower summer flows [4]. The summers will be dryer but, regardless of season, the predictions indicate an increase of intensive rainfall and extreme flows [3,4]. The changes will lead to increased erosion rate and also other impacts that will affect the slope stability such as water level changes and changes in vegetation, e.g. [1,5]. One of the conclusions in the Swedish climate and vulnerability inquiry [6] was that the increased risks related to erosion and landslides will increase in the coming decades to such an extent that adaptation measures are needed to be taken already today.

Land slide events, such as other natural hazard events, are not disasters per se but the consequences may besevere and even disastrous [7-13]. In technical risk analysis, the landslide risk is therefore referred as a function of the probability of a landslide event and the consequences thereof and the risk can be mitigated by either reducing the probability, the consequences of an event or both, e.g. [8,14-17].

For a well-functioning risk mitigation strategy, the risk mitigation requirement needs to be determined. The risk mitigation requirement is a function of the relation between the results of a risk assessment and the accepted level of risk, e.g. [8,17,18]. The risk assessment is done by assessing the probabilities of the unwanted events, e.g. landslides, and assessing the consequences of each event.

The causes of landslides are site specific and complex. The cause can be both primary (caused by factors that are long lasting and inherent in the constituent rock and the soil) and caused by triggers (factors that are varying or very short lived), e.g. [13]. Due to the wide range of landslide processes, the annual probability and the extension of a potential landslide are difficult to assess and the environmental and societal context is crucial for determining the specific effects on the vulnerability as pointed out by Andersson-Berry & King [19] and Roberts et al. [17]. The examples of quantitative assessments of landslide probability reported in the literature have been sited specific and based on statistical information combined with ground properties and parameter uncertainty analysis [11,20-25].

The consequence is a function of the amount of elements at risk (the numbers exposed) and the vulnerability of the affected element or system at risk [22]. Vulnerability is the degree to which a system is susceptible to, or unable to cope with the situation. It can be defined as a function of the character, magnitude and rate of the event to which a system is exposed, its sensitivity, and its adaptive capacity [5,26], or in other words the exposure to the event (perturbation and stresses) combined with the sensitivity and resilience and adaptive capacity of the system [13,14,16,26-28]. Even though the landslide probability is complex to determine the vulnerability, it is even more complex that there is no generic method for assessing the sensitivity, resilience and adaptive capacity for all types of elements at risk. For some impacts, quantitative assessment is possible while for others only qualitative assessments are preferable [8].

Landslide risk (R) is, as described above, referred to as a function of probability and the consequence of a landslide. Recently it often also includes the costs related to the event, i.e. R = H*V*E where, H is the land hazard (landslide) probability, V is the physical vulnerability in case of an event and E is the cost of the particular elements at risk, e.g. [23]. To involve the cost in the risk, assessment function is relevant when assessing the costs and benefits of risk mitigation. As pointed out by van Westen et al. [29] and Crovelli and Coe [23], although the risk equation is simple, it is difficult to be applied in practise due to difficulties in determining H, V and E. Yet, the ability to forecast losses is critical for effective landslide risk management.

Despite the difficulties, cost benefit analyses (CBA) have recently been performed on numerous risk management options, i.e. physical measures [30], landslide susceptibility mapping [20], warning systems and physical measures, the efficiency of incorporating risk mitigation in local comprehensive plans [9] and risk reduction programs [31]. Crovelli and Coe [23] estimated the number of future landslides for a given time period and the economic losses from those landslides, based on historical landslide costs (losses) to assess the expected costs in the San Francisco bay region. The results can be, and were, applied for future prognosis but also to relate the costs to other regions. Similar future cost estimates have been done as presented in [10,20,22,32-36]. In such analyses, the consequence assessments are generally based on loss of life and material (buildings) including reconstruction and renovation costs. In some assessments, losses of income due to delays or closures are included [37]. In general, sensitivity and time-dependent exposure are not taken into account. In some countries, like Sweden, there are not enough data that can be applied for statistically based landslide prognoses [38]. Instead the probability needs to be assessed based on field measurements of the current situation combined with slope stability calculations. On the other hand, Sweden has a wealth of data when it comes to consequence analysis.

The aim of this paper is to present a quantitative landslide risk assessment methodology developed for semiregional scale such as a large run-off area and taking into account a wide range of consequences. The method has been applied for assessing the risk, in the Göta river valley, a large river system in one of Sweden’s most landslide-prominent areas [39,40]. The work was initiated by the Swedish Government in response to the results of the Swedish climate and vulnerability inquiry [6]. The risk assessment is in Swedish referred to as Göta älvutredningen [41].

2. METHOD

2.1. The Study Area

The Göta river valley runs from Lake Vänern in thenorth to Göteborg in the south (Figure 1). The valley is one of the most frequent landslide valleys in Swede [39,40] with a number of landslides occurring every year. In general the landslides are fairly small, shallow and caused by erosion. Larger landslides have also occurred in the river valley resulting in human casualties and extensive property damage [39].

Geologically the prerequisites for landslides formed during and following the latest glaciation period, whendeep layers of clay formed in the river valley which wassubmerged in the sea during this period. Since the material was mainly deposited in a marine environment, quick clay is common in the area. Quick clay is a soil with high water content and weak bindings between the clay particles. Vibrations, or a small initial landslide, can cause a quick clay layer to collapse and liquefy, resulting in a large rapid landslide [42]. The river flow is regulated by the power dam in Lilla Edet and the so called Vänern regulation, i.e. the regulation of the power dam tapping from Lake Vänern, which has been applied since 1937. Already in 1963 Sundborg and Norrman mentioned that

Figure 1. The Göta river valley (study area marked in grey). © SGI, Lantmäteriet, Geodatasamverkan.

the regulation caused increased erosion and frequency of river beachfront landslides [41,43]. Climate change is expected to increase the frequency of extreme flows (high and low) from Lake Vänern further contributing to increased erosion [44,45].

The Göta river valley has a long history of anthropogenic activities such as settlements, shipping, harbours, industry, contaminated soil and infrastructure including large roads and railroads, e.g. [46]. Consequently the consequences of a large landslide can be severe.

2.2. Methodology for Quantitative Landslide Risk Assessment

2.2.1. Landslide Probability

The landslide probability was assessed based on slopestability factor estimates along the investigated area. The slope stability was calculated applying an overall security profile in accordance with the national code for practice instructions [47] by Slope/W 2007:7:13 (or later versions) method Morgenstern-Prices with “optimized” sliding surface, or the software GeoSuiteStabilitet 4 (or later versions) method Beast 2003. In all calculations hard crust and half way water filled cracks was assumed. Road and railroad loads were based on TK Geo [48] and building loads were assumed 10 kPa per estate floor. The calculations were done for both drained and undrained analyses, and the lowest resulting safety factor was used. Water levels were set as the lowest allowed level based on tapping and sinking limits of the river system and the average low sea water level as described in more detail in [49]. The slope stability calculations were based on the following information:

• Compilation on geology and stratigraphy in the valley based on recent investigations [50-53].

• GIS based terrain model in SWEREF99TM, RH 2000 [54] based on recent air and sea based laser scanning by Vattenfall ABin 2007, complemented by multibeam echosounding of the river bed within this investigation [55].

• Geotechnical field measurements for in total 240 sections perpendicular to the river. Along each section up to 6 bore holes have been investigated regarding: Total pressure soundings (664), CPT (804), soil-rock soundings (12), impact sounding (21); remoulded sample (333); vane shear testing (379); piston sampling (291); pore water pressure (557 in 70 stations). In total ca 2500 geotechnical analyses were done, and were complemented with information from previous geotechnical investigations in the area.

• Laboratory investigations of field samples from the investigated sections were done in accordance with SIS-CEN ISO/TS 17892-nr including earth measurements and type of soil estimates (3900), bulk density (3600), water content (3500), cone liquid limit (3100), cone tests undisturbed soils (3000), remoulded soils (2200), CRS (755), direct shear tests (106) and triaxial tests (7).

• Current ground water conditions based on field measurements (34 stations) combined with a compilation of previous investigations [56].

• Current erosion in the river was assessed based on investigations in the area [49].

Based on the information described above the landslide probability factor, Pf, was calculated. Pf was calculated for in total 80 sections in 39 different type areas (based on topography and geological conditions including geometry, undrained shear strength and soil density). The specific conditions for each type area were based on a previous classification by Millet [57]. The landslideprobability factor was calculated by First Order Reliability Method (FORM) [58]. The marginal to slope failure, the so called safety marginal, M, which is the difference between the load capacity and the load, is central in this method. If M > 0 the load capacity is larger than the load effect and for M < 0 the load capacity is less than the load effect. The landslide probability factor is the probability for slope failure, Equation (1):

. (1)

The safety marginal, as applied here, was calculated as a function of the safety factor, F, Equation (2):

. (2)

F was calculated in accordance with Janbu [59] based on the model error, ƞmodel, average shear strength along the sliding surface, c, the stabilitynumber, N, and the driving force pressure, Pd. The model error is applied to compensate for the deviation caused by the idealization of the modeled slope [59]. The stability numbers were taken from Janbu [59] and Alén [60]. The driving force pressure (Pd) was calculated by Equation (3):

. (3)

where γ is the average pressure caused by the soil layers above the slope foot, H is the vertical distance between the slope crest and foot, qe is the equivalent surface load behind the slope top, γw is the pressure caused by the water in the water course and Hw is the distance between the water level in the water course and the slope foot. Including Equation (2) into Equation (1) results in Equation (4):

(4)

Which, when taking into account the variation of data information, can be expressed as Equation (5):

, (5)

where Φ is the standardized normal distribution and β is the safety index which is calculated based on the average safety factor (μF) and the safety factor variation coefficient (VF) in accordance with Equation (6):

, (6)

where VF ≈ the standard deviation for the safety marginal, σlnF.

The results from the calculations are the landslide probability factor Pf and the safety factor, F. These parameters were used to express the landslide probability in five classes, S1 - S5, in accordance with previous Swedish classification system and previous risk classification system for the Göta älv river system [61] synchronised with Eurocode, i.e. S3 corresponds to Eurocode’s lowest accepted security level, as described in detail in Göta älvutredningen [41]. The resulting classification system is presented in Figure 2.

The probability factors and safety factors from the type sections were then applied on the additionally investigated sections, ca 200, to estimate the landslide probability class for each of those sections based on the estimated slope stability.

To include the potential extension of a landslide, thequick clay distribution in the investigated area was identified by total pressure and CPT as described by Möller and Bergdahl [62] and recently applied by Rankka et al. (2004). The Swedish definition of quick clay [42,63], i.e. sensitivity (St) > 50 and remoulded shear strength (τR) < 0.4 KPa, was applied.

2.2.2. Consequences

The development of a method for consequence classification has previously been described [46]. The starting point of the consequence analysis was landslide consequence identification, followed by an inventory based on interviews, previous investigations and events. The inventory also included GIS based data on vulnerable elements including: Human life and personal injuries; Property; Potential income losses (industry and other commercial enterprises); Environmentally hazardous activities, i.e. elements under the emission register EMIR and Seveso classified elements; Contaminated land areas; Infrastructure of national importance: roads, railroads, seatraffic, heating, electricity, communication (tele, IT) and water supplies and sewage systems; Natural and cultural environments. The results of the inventories of each sec-

Figure 2. Correlation between probability failure, Pf, and average safety factor, F, for different safety factor variation coefficient, VF values and the five probability classes, S1 - S5 (modified from Berggren et al. [58]).

tor were collected in a GIS raster layer with resolution 100 × 100 m covering the area under investigation [46].

As the consequence in case of an event do not only depend on the number of elements that can be exposed but also the exposure at the time and the sensitivity and adaptive capacity of the system exposed, exposure factors and vulnerability factors were estimated. The exposure factor varies from 0 (no exposure) and 1 (full exposure). The landslide exposure factor for stationary objects such as buildings, roads and railroads, contaminated land and cultural artefacts was set to 1. The exposure factor for human life depend on how people divide their time between work, home, school, etc. at different physical addresses. Human exposure was estimated by applying the results from a time usage investigation for year 2000/01 [64] resulting in an annual average exposure factor of 0.69 for people in their homes. Exposure factors at work and in schools were estimated to an annual average of 0.24 and 0.14 respectively [46].

The vulnerability factor was applied to describe the combined effects of the sensitivity and adaptive capacity. The vulnerability factor for a property can be described as the degree of loss or of damages, and for people the mortality (probability of death) caused by the landslide. The vulnerability factor ranges from 0 (no loss) to 1 (total loss) for primary landslide impacts. The vulnerability factor for human life was determined from previous landslides in Sweden and Norway, i.e. the Swedish Natural Hazards Information System [65] and the Norwegian database Skrednett [66] and was expressed as a normal distribution with μ = 0.16 and σ = 0.09 0.16 ± 0.09 [46]. The vulnerability related to a landslide on contaminated land could, for example, be estimated based on the changed risks for human or environment related to the changed exposure of contaminants in the soil both in a short and long time perspective. There is, however, too scarce information for such an estimate. Therefore an alternative approach was applied, in which the impacts were valued using the extra costs related to activities to make the landslide area safe to work on after a landslide event. Similarly, flat values of potential extra costs wereapplied for environmentally hazardous activities. For roads, railroads and shipping the impacts on delays, traffic closure and changes of routes were included, as well as reconstruction costs. Costs related to traffic delays and impacts from redirecting road and railroad traffic were estimated through simulations applying different traffic models, i.e. Sampers1 and Eva2, and the results were implemented in the model applying flat rate values. The vulnerability factor, for a building or any other stationary object, was set to 1 [46].

The methods for monetary valuation are summarized in Table 1. Details can be found in Andersson-Sköld and Falemo [46]. Once the potential consequence had been estimated based on the inventory, exposure and vulnerability factors, i.e. Consequence = Inventory*Exposure factor*Vulnerability factor*Monetary value, the consequences were calculated in monetary terms foreach sector and 100 × 100 m grid square GIS of the investigated area. The raster layers were summed up in a GIS raster calculation resulting in total estimated monetaryvalue of the consequences in each grid square. The raster cells were then summed up in rows perpendicular to the river as the basis for creating consequence class-isolines as shown in Figure 3. Five consequence classes were

Table 1. Methods used for socio-economic valuation of potential landslide consequences [46].

Figure 3. Examples of a resulting consequence map (left) and consequence class iso-lines (right, red lines). The consequences are shown as five classes and the total monetary value within each grid cell: grey (EUR65 million). © SGI, Lantmäteriet, Geodatasamverkan.

used to map the consequences, K1 - K5 (Table 2).

As the monetary valuation is site/context specific the values represented within each of the classes need to be decided among stakeholders/experts with knowledge about the specific investigated area. In this investigation the consequence class values were based on an existing qualitatively based, consequence classification system previously applied in the area [39,61,67]. The resulting landslide consequence map, displaying the resulting consequence classes, illustrates the estimated loss should that entire cell be directly affected by a landslide [46].

2.2.3. Risk Classes and Risk Mapping

The consequence map was then over-layered with the probability map and combined into risk classes. A risk matrix was used to classify consequences and probabilities applying the five classes of each, i.e. S1 - S5 and K1 - K5 as summarised in Table 2.

The risk is presented as pair of values, i.e. the failure probability class and the consequence class, in a matrix system (Figure 4). In order to facilitate the readability into a resulting risk map, the resulting risk was divided into three classes, i.e. low risk, medium and uncertain risk, and high risk, coloured yellow, orange and red respectively (Figure 4).

2.2.4. Time Dependence

The slope stability changes with time due to continuous change in the landslide prerequisites (e.g. erosion and changes in precipitation which alter the river surface level and the pore water pressure). These changes in prerequisites were estimated through scenario based calculations of the safety factor. The scenarios included changes in erosion, water levels and pore-water pressure. Future groundwater scenarios were achieved by combining the current groundwater situation with future climate parameters [68]. The surface water, i.e. the river water levels and river water flow for future conditions were based on scenario calculations by SMHI [45] including the expected tapping from Lake Vänern to Göta River [44]. Erosion changes were estimated based on available erosion measurements; water flow scenarios under conditions where current tapping regulation is applied; and for conditions with higher tapping volumes, by applying expert judgments by hydrologists, experts in sedimentology and geotechnical experts. The estimates of the erosion are, consequently, very uncertain and include rough assumptions and rough parameterization.

The results from the erosion calculations shall therefore only be used as indicators of potential magnitudes of changes. In reality the landslide prerequisites may change also due to changes in land use or other man made load changes, such changes have not been taken into account. When assessing the risk over a longer period, such as 100 years, in principle not only the landslide probability but also the time dependence of the consequences shall be included. There are several uncertainties in the changes related to the consequences, including both the social development in the area where the uncertainties of the demographic changes, industrial

Table 2. Landslide probability and consequence class intervals.

*The values within each of the classes need to be decided among the stakeholders/experts for each site specific area.

Figure 4. Risk matrix based on landslide probability and consequence classes.

development and socio-economic development are large as are the variations within the investigated region, as are the uncertainties and variations of the current and forthcoming activities and ambition levels regarding physical planning to mitigate the risks etc. Spatial planning in the time perspective of 100 years is expressed in visions rather than in comprehensive plans. Further, the consequence costs estimated in this investigation are not complete due to lack of available knowledge and information, such as costs and vulnerability factors related to human injuries and general knowledge on landslide impacts on the natural environment [46]. Due to the large uncertainties, the time dependence, and consequently any discount rates of consequence costs have not been estimated in this investigation.

3. RESULTS

The major result of the work described here is the development of a methodology that can be applied for landslide risk assessment and mapping anywhere where sufficient GIS data is available. The major steps involved in the methodology are described in Figure 5.

The methodology presented in Figure 5 is generic. For each of the individual steps, e.g. in the landslide probability assessment, several methods can be applied. The basis for the assessment can be based on available information or on new inventories and field measurements. In this paper the results from applying the above described methods in the Göta river valley are presented. The resulting risk map is shown in Figure 6. The map is meant to be used as a basis for risk management in the area on local to regional scale. The consequences have a resolution of one hectare and the slope stability analyses were done for sections spaced between 400 and 1500 m, and the map accordingly has to be used for such large scale risk management. For example the high risk class areas (red) need to be investigated in more detail prior to finalizing strategies of physical risk mitigation actions in specific areas. Medium risk class areas (orange) also need to be investigated further. The inclusion of the pair of values can be used to identify whether the more detailed investigations shall be focused on the elements at

Figure 5. Stepwise model for GIS based risk map on semi regional scale (river valley scale).

risk, the slope stability assessment, or both.

3.1. Climate Change

The map does not only include information on the current situation but also on expected impacts of climate change. The main impact is due to changes in erosion and thereby the geometric changes of the slopes in and above the river shore line. The erosion depends on the water flow and the water levels, which in turn depend on the maximum tapping and its duration. In the water flow and erosion simulations different future tapping scenarios were included. The maximum tapping is currently regulated to 1030 m3/s, while in the future either the duration of such tapping will need to increase or the regulation needs to be changed to allow higher maximum tapping. The maximum tapping and potential duration for different potential tapping scenarios is presented in Table 3.

The erosion under current conditions and regulations is 0.05 m per year in most parts of the river, with a few exceptions with up to 0.15 m per year.

The erosion under current regulation but in a future climate is expected to be of the same annual mean as today, which is 0.4 to 0.5 m in the northern parts of the river and 1.0 to 1.5 m in the other parts over the next 90 years [49]. For the tapping scenarios 1 - 3, shown in  Table 3, the erosion will increase, the higher the water flow the higher the erosion. As the most likely tapping will be a mix of the different scenarios the average increase in erosion for the tapping alternatives 1 - 3 is 0.8 to 2 m in the northern parts of the river and up to be-

Figure 6. Example of risk map (from GÄU, 2013, part 3 [41]). © SGI, Lantmäteriet, Geodatasamverkan.

Table 3. Maximum tapping and duration under current conditions3 and simulated for different tapping scenarios [45].

tween 2 - 3 meters in the southern parts of the river the nearest 90 years. The details regarding the erosion estimates are further described in Rydell et al. [49].

The impact of erosion on slope stability is estimated to be marginal for areas where the erosion is less than 1m in 100 years. In areas where the erosion is expected to be 1.5 m the slope is expected to move 4 m inland compared to current situation and the safety factor will be reduced by 4%. For the areas and tapping scenarios where the erosion is simulated to be 2 - 3 meters in 100 years, the safety factor will be reduced to 14% on average. In Figure 7 all sections where the slope stability would be affected by 1 m and 3 m erosion are presented along with the impact on the stability class.

As can be seen from Figure 7 the impact of erosion is more pronounced in areas where there currently is a low landslide probability compared to where it is high and in general the impact is higher for the higher (3 m) erosion. The potential impact of erosion in the investigated area is shown on the risk map by a raster marking when there may be potential medium or large impacts on the slope stability the nearest 90 years due to climate change and related needs of changed tapping regulations (Figure 6).

3.2. Cost Assessment

The result from the risk map was used to estimate thevalue that would be affected in case of a landslide event in the entire area of the investigation (excluding rockyground). The total value was estimated to around EUR17 billion (150 BSEK). The corresponding value within high risk class areas (red in Figures 4 and 6) were calculated to ca EUR 900 million (8 BSEK) under current conditions. The high risk classed areas (red) will expand due to the shift to higher landslide probability classes in many parts of the investigated area as a consequence of climate change causing higher water flows and thereby increased erosion.

3.3. Mitigation Measures

In areas with high risk (red), risk mitigation measures were suggested by expertise within SGI and a rough cost assessment of the suggested measures was done based on current market prices and experience. The mitigation measures ranged from erosion protection, excavations and embankment to increased maintenance and monitoring and more detailed geotechnical investigations [41]. The estimated cost for suggested measures to decrease the landslide probability by 20% (compared to current landslide probability) in the high risk class areas (red) was EUR550 - 700 million (5 - 6 BSEK) with operational costs around EUR10 million (900 MSEK) per year.

The cost hence is less, but in the same range, as the protected value in the protected areas. The difference would be higher if all damages such as personal injuries, personal and municipal administration costs, affection values, cost of stress and sorrow etc. had been accounted for.

3.4. Communication with Stakeholders

The resulting risk map was communicated with the key stake holders in the region such as municipalities

Figure 7. Impact on slope stability class due to increased erosion compared to current conditions.

along the river, the regional road and railroad administration, the county administration board, the waterpower industry and the Swedish maritime administration and on national level including for example Swedish Civil Contingencies Agency, the Ministry of the Environment and the Ministry of Health and Social Affairs. Stakeholder communication was initiated by the start of the project and with occasional communication through the course of the project. By the end of the project, the final results were presented and discussed in stakeholder workshops and seminars. The final maps were made available to stakeholders and the public in March 2012. SGI experts had in advance booked meetings with all municipalities along the river to discuss the impacts of the results within each municipality. The maps have been applied as discussion and decision bases in the municipalities, on county and on governmental level to estimate the needs of risk mitigation and to make priorities.

4. DISCUSSION

The main result of the project presented here was a step wise method for landslide risk mapping including landslide probability assessment and monetary valuation of the potential consequences. In the landslide probability assessment applied in the Göta river valley, the upto-now most thorough field investigation of the area was combined with traditional slope stability analyses, model estimates of current and future erosion and water flows and levels, quick clay mapping and quick modelling to assess the current and future stability and the potential extension of a landslide. The consequence analysis covered more elements at risk than normally included in landslide risk assessments, taking into account both direct and indirect consequences. The reason for such a thorough analysis was the need of more detailed understanding of the risks in the area both under current conditions and taking into account climate change. For efficient mitigation strategies selected areas also need even more detailed investigations and assessments. In such cases the generic method shown in Figure 5 can, and will be used, but some steps already are described well enough for the purpose. We accordingly recommend the stepwise method to be used in an iterative process focusing on the needs at the moment and aiming to find which areas need further and deeper assessment.

Despite being regarded as a detailed and thorough assessment the results are based on many rough estimates and assumptions, and many uncertainties and unknowns still exist. Among the most important uncertainties are the estimates of current and future erosion and the quick clay distribution which are needed to assess the current and future stability and the potential extension of a landslide. For more accurate assessments several knowledge gaps needs to be filled regarding sediment transport and erosion processes for the different soils in the river. The mapping of quick clay is costly and methods and increased knowledge for faster and cheaper mapping is requested for more thorough estimates.

There are also large uncertainties in the values of the consequences and for some consequences, such as the nature environment, not even the magnitude and whether it is positive or negative impact can be estimated [46]. Of special concern are landslides on contaminated land or enterprises that may cause emissions of contaminants to air, soil and water as there is not enough information to describe the short and long terms impacts, and therefore they cannot even be evaluated. In the same way the impact on biodiversity and various biotopes of a landslide may be both positive and negative and both are in general impossible to quantify with current knowledge and thereby not available for monetary valuation. Human injuries are omitted and costs related to increased administration and soft impacts are also omitted.

The resulting map has been communicated and applied in the landslide risk management by stakeholders and municipalities in the investigated area and despite the uncertainties the resulting map is regarded as very useful and important among the stakeholders. The application of GIS, and that the resulting map is available on the web, is also highly appreciated among the stakeholders. This is in agreement with previous results on the perception of risk mapping and the communication with experts on how to interpret their information [38,69]. The map is useful as it indicates landslide prone areas with high population densities and important vulnerable infrastructure areas that shall be further investigated and regarded for further risk reduction activities. In areas identified as high risk areas, the consequence values are higher, but of the same magnitude as the costs estimated for preventive measures in the same areas. This indicates that preventive measures are beneficial, but also that more detailed site specific analyses are needed prior to identifying and implementing the most cost efficient risk reduction strategies. One of the major reasons for the relative low monetary value in high risk areas are previous landslide prevention measures taken. An example is EKA Nobel where in 1994 land reinforcements for EUR8 million (70 MSEK) were undertaken. Another example is active land use planning to reduce the vulnerability in the river valley [38]. This is in agreement with an iterative process, i.e. that the obvious needs can be treated without as thorough investigation as the one presented here, while in cases where the value of expected losses are close to the costs of risk mitigation measures more detailed analyses are needed. The generic step wise method and GIS based information and data is then easy to apply and combine with the information of the risk map for the larger region, thereby providing information that can be used to develop a more holistic risk management approach and strategy in the entire investigated area.

5. CONCLUSIONS

Risk maps are important tools in communication with stakeholders that may be impacted by potential landslides and/or are responsible for landslide risk management and prevention [38,69].

The level of detail of the risk map must depend on the scale investigated, and to achieve the most cost effective information level, a stepwise method is recommended to be used in an iterative process focusing on the needs at the moment and aiming to find which areas need further and deeper assessment. The method presented here (Figure 5) has been found applicable and useful in the preventive work in the investigated area.

There are many uncertainties related to understanding the risks both under current climate conditions and, not least, taking into account climate change.

There is a need for a deeper understanding of the process that may cause serious exposure, and other consequences, to health and environment due to landslides in order to make better quantitative assessments of the consequences.

In general, the understanding of the consequences could be increased if there were more systematically available data with regard to previous events including the number of people exposed and number of injuries in addition to the number of fatalities most commonly presented.

The knowledge on social and individual levels, and the social costs, related to landslides and other severe hazards needs to be better understood.

REFERENCES

  1. IPCC (2007) Climate change 2007: Impacts, adaptation and vulnerability. Parry, M.L., Ed., Cambridge University Press, New York.
  2. Andréasson, J., Bergström, S., Carlsson, B., Graham, P. and Lindström, G. (2004) Hydrological change: Climate change impact simulations for Sweden. A Journal of the Human Environment, 33, 228- 234.
  3. Lawrence, D., Graham, L.P., den Besten, J., Andréasson, J., Bergström, S. and Engen-Skaugen, T. (2012) Climate change impacts and uncertainties in flood risk management: Examples from the North Sea Region. A Report of Working Group 1, SAWA Interreg IVB Project, 2012, Report No. 05.
  4. Olsson, J., Yang, W., Graham, L.P., Rosberg, J. and Andréasson, J. (2011) Using an ensemble of climate projections for simulating recent and near-future hydrological change to Lake Vänern in Sweden. Tellus A, 63, 126-137. http://dx.doi.org/10.1111/j.1600-0870.2010.00476.x
  5. IPCC (2001) Climate change 2001: The scientific basis. contribution of working group I to the third assessment report of the intergovernmental panel on climate change. Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K. and Johnson, C.A., Eds., Cambridge University Press, Cambridge, United Kingdom and New York, 881.
  6. SOU 2007:60. Sweden facing climate change: Threats and opportunities. Final Report, Fritze, Stockholm
  7. Berke, P.R. (1998) Reducing natural hazard risks through state growth management. Journal of the American Planning Association, 64, 76-87.
  8. Dai, F.C., Lee, C.F. and Ngai, Y.Y. (2002) Landslide risk assessment and management: An overview. Engineering Geology, 64, 65-87. http://dx.doi.org/10.1016/S0013-7952(01)00093-X
  9. Srivastava, R. and Laurian, L. (2006) Natural hazard mitigation in local comprehensive plans: The case of flood, ildfire and drought planning in Arizona. Disaster Prevention and Management, 15, 461-483. http://dx.doi.org/10.1108/09653560610669936
  10. Sterlacchini, S., Frigerio, P. and Giacomelli, M. (2007) Brambilla, landslide risk analysis: A multi-disciplinary methodological approach. Natural Hazards and Earth System Sciences, 7, 657-675. http://dx.doi.org/10.5194/nhess-7-657-2007
  11. Friele, P., Jakob, M. and Clague, J. (2008) Hazard and risk from large landslides from Mount Meager volcano, British Columbia, Canada. Georisk, 2, 48-64. http://dx.doi.org/10.1080/17499510801958711
  12. Holcombe, E. and Anderson, M. (2010) Tackling landslide risk: Helping land use policy to reflect unplanned housing realities in the Eastern Caribbean. Land Use Policy, 27, 798-800. http://dx.doi.org/10.1016/j.landusepol.2009.10.013
  13. Singh, A.K. (2010) Landslide management: Concept and philosophy. Disaster Prevention and Management, 19, 119-134. http://dx.doi.org/10.1016/j.landusepol.2009.10.013
  14. Brooks, N. (2003) Vulnerability, risk and adaptation: A conceptual framework, tyndall centre for climate change research. University of East Anglia, Norwich.
  15. Sarewitz, D., Pielke Jr., R. and Keykhah, M. (2003) Vulnerability and risk: Some thoughts from a political and policy perspective. Risk Analysis, 23, 805-810. http://dx.doi.org/10.1111/1539-6924.00357
  16. Füssel, H.M. and Klein, R.J. (2006) Climate change vulnerability assessments: An evolution of conceptual thinking. Climate Change, 75, 301-329. http://dx.doi.org/10.1007/s10584-006-0329-3
  17. Roberts, N.J., Nadim, F. and Kalsnes, B. (2009) Quantification of vulnerability to natural hazards. Georisk, 3, 164-173. http://dx.doi.org/10.1080/17499510902788850
  18. King, D. (2008) Reducing hazard vulnerability through local government engagement and action. Natural Hazards, 47, 497-508. http://dx.doi.org/10.1007/s11069-008-9235-5
  19. Anderson-Berry, L. and King, D. (2005) Mitigation of the impact of tropical cyclones in Northern Australia through ommunity capacity enhancement. Mitigation and Adaptation Strategies for Global Change, 10, 367-392. http://dx.doi.org/10.1007/s11027-005-0052-9
  20. Remondo, J., González, A., Díaz de Terán, J.R., Cendrero, A., Fabbri, A. and Chung, C.J.F. (2003) Validationof landslide susceptibility maps; examples and applications from a case study in northern Spain. Natural Hazards, 30, 437-449. http://dx.doi.org/10.1023/B:NHAZ.0000007201.80743.fc
  21. Uzielli, M. and Lacasse, S. (2007) Scenario-based probabilistic estimation of direct loss for geohazards. Georisk, 1, 142-154.
  22. Cassidy, M.J., Uzielli, M. and Lacasse, S. (2008) Probability risk assessment of landslides: A case study at Finneidfjord. Canadian Geotechnical Journal, 45, 1250- 1267. http://dx.doi.org/10.1139/T08-055
  23. Crovelli, R.A. and Coe, J.A. (2009) Probabilistic estimation of numbers and costs of future landslides in the San Francisco Bay region. Georisk, 3, 206-223. http://dx.doi.org/10.1080/17499510802713123
  24. Ho, K.K.S. and Ko, F.W.Y. (2009) Application of quantified risk analysis in landslide risk management practice: Hong Kong experience. Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 3, 134-146. http://dx.doi.org/10.1080/17499510902873074
  25. Kanungo, D.P., Arora, M.K., Sarkar, S. and Gupta, R.P. (2009) A fuzzy set based approach for integration of thematic maps for landslide susceptibility zonation. Georisk, 3, 30-43.
  26. Rekacewicz, P. (2005) Vital climate change graphics, UNEP/GRID-arendal maps and graphics library. http:maps.grida.no/go/graphic/sensitivity-adaptability-and-vulnerability
  27. Turner, B.L., Kasperson, R.E., Matsone, P.A., McCarthy, J.J., Corell, R.W., Christensen, L., Eckley, N., Kasperson, J.X., Luerse, A., Martello, M.L., Polsk, C., Pulsipher, A. and Schiller, A. (2003) A framework for vulnerability analysis in sustainability science. Proceedings of the National Academy of Sciences of the United States od America, 100, 8074-8079.
  28. Li, Z., Farrokh, N., Hongwei, H., Uzielli, M. and Lacasse, S. (2010) Quantitative vulnerability estimation for scenariobased landslide hazards. Landslides, 7, 125-134. http://dx.doi.org/10.1007/s10346-009-0190-3
  29. Van Westen, C.J., Castellanos, E. and Kuriakose, S.L. (2008). Spatial data for landslide susceptibility, hazard, and vulnerability assessment: An overview. Engineering Geology, 102, 112-131. http://dx.doi.org/10.1016/j.enggeo.2008.03.010
  30. Schuster, R.L. and Highland, L.M. (2007) The third hans cloos lecture. Urban landslides: Socioeconomic impacts and overview of mitigative strategies. Bulletin of Engineering Geology and the Environment, 66, 1-27. http://dx.doi.org/10.1007/s10064-006-0080-z
  31. Anderson, M.G., Holcombe, E., Esquivel, M., Toro, J. and Ghesquiere, F. (2010) The efficacy of a programme of landslide risk reduction in areas of unplanned housing in the Eastern Caribbean. Environmental Management, 45, 807-821. http://dx.doi.org/10.1007/s00267-010-9431-4
  32. Plate, E.J. (2007) Flood risk management for setting priorities in decision making, in Extreme Hydrological Events: New Concpepts for Security. In: Vasiliev, O.F., van Gelder, P.H.A.J.M., Plate, E.J. and Bolgov, M.V., Eds., Extreme Hydrological Events: New Concpepts for Security, 78, 21-44.
  33. Uzielli, M., Farrokh, N., Lacasse, S. and Kaynia, A.M. (2008) A conceptual framework for quantitative estimation of physical vulnerability to landslides. Engineering Geology, 102, 251-256. http://dx.doi.org/10.1016/j.enggeo.2008.03.011
  34. Hinkel, J., Nicholls, R.J., Vafeidis, A.T., Tol, R.S.J. and Avagianou, T. (2010) Assessing risk of and adaptation to sea-level rise in the European Union: An application of DIVA. Mitigation and Adaptation Strategies for Global Change, 15, 703-719. http://dx.doi.org/10.1007/s11027-010-9237-y
  35. Zeng, J., Zhu, Z.Y., Zhang, J.L., Ouyang, T.P., Qiu, S.F., Zou, Y. and Zeng, T. (2012) Social vulnerability assessment of natural hazards on county-scale using high spatial resolution satellite imagery: A case study in the Luogang district of Guangzhou, South China. Environmental Earth Sciences, 65, 173-182. http://dx.doi.org/10.1007/s12665-011-1079-8
  36. Holcombe, E., Smith, S., Wright, E. and Anderson, M.G. (2012) An integrated approach for evaluating the effectiveness of landslide risk reduction in unplanned communities in the Caribbean. Natural Hazards, 61, 351-385. http://dx.doi.org/10.1007/s11069-011-9920-7
  37. Eder, S., Poscher, G. and Prager, C. (2006) Risk analyses and risk management—Slope instabilities in Alpine environments. ECI Geohazards Volume P07-Technical, Economical and Social Risk Evaluation.
  38. Andersson-Sköld, Y., Bergman, R., Johansson, M., Persson, E. and Nyberg, L. (2013) Landslide risk management: A brief overview and example from Sweden of current situation and climate change. International Journal of Disaster Risk Reduction, 3, 44-61. http://dx.doi.org/10.1016/j.ijdrr.2012.11.002
  39. Hultén, C., Andersson-Sköld, Y., Ottosson, E., Edstam, T. and Johansson, Å. (2007) Case studies of landslide risk due to climate change in Sweden. Proceedings of International Geotechnical Conference on Climate Change and Landslides, Ventnor, 21-24 May 2007, 149-157. http://dx.doi.org/10.1201/NOE0415443180.ch19
  40. Jonsson, A., Hjerpe, M., Andersson-Sköld, Y., Glaas, E., André, K. and Simonsson, L. (2012) Cities’ capacity to manage climate vulnerability: Experiences from participatory vulnerability assessments in the lower Göta Älv Catchment, Sweden. Local Environment, 17, 735-750. http://dx.doi.org/10.1080/13549839.2012.685880
  41. GÄU (2012) Skredrisker i Göta älvdalen i ett förändrat klimat. Swedish Geotechnical Institute, Slutrapport. Göta älvutredningen 2009-2011. http://www.swedgeo.se/templates/SGIStandardPage____2634.aspx?epslanguage=SV )
  42. Andersson-Sköld, Y., Torrance, J.K., Lind, B., Odén, K., Stevens, R.L. and Rankka, K. (2005) Quick clay: A case study of chemical perspective in Southwest Sweden. Engineering Geology, 82, 107-118. http://dx.doi.org/10.1016/j.enggeo.2005.09.014
  43. Sundborg, Å. and Norrman, J. (1963) Göta älv—Hydrlologi och morfologi med särskild hänsyn till erosionprocesserna. Svergiges Geologiska Undersökning, SGU. Rapport Ca 43.
  44. Bergström, S., Andréasson, J., Asp, M., Caldarulo, L., German, J., Lindahl, S., Losjö, K. and Stensen, B. (2010) Fördjupad studie rörande översvämningsriskerna för Vänern—Slutrapport. SMHI Rapport nr 2010-85.
  45. Bergström, S., Andréasson, J., Losjö, K., Stensen, B. and Wern, L. (2011) Hydrologiska och meteorologiska förhållanden i Göta älvdalen—Slutrapport. Statens geotekniska institut, SGI. Göta älvutredningen, GÄU. Delrapport 27. Linköping. Även publicerad som SMHI Rapport nr 2010-81.
  46. Andersson-Sköld, Y. and Falemo, S. (2013) Ex ante natural hazard consequence and vulnerability analysis for local to regional scale—Method development focusing on landslide events. Manuscript submitted Natural Hazards.
  47. Skredkommissionen (1995) Anvisningar för släntstabilitetsutredningar, ingenjörsvetenskapsakademien, IVA. Skredkommissionen. Rapport 3:95, Linköping.
  48. Vägverket (2009) TK Geoär en teknisk beskrivning som anger Vägverkets och Banverkets krav och råd vid nybyggnad och förbättring av geokonstruktioner. Vägverket, VV. Publikation 2009:46/Banverket, BV. BVS 1585.001, Borlänge.
  49. Rydell, B., Persson, H. and Blied, L. (2011) Klimateffekt på vattennivå, erosion och grundvattenförhållanden i Göta älv. Underlag till styrdokument 99ST003. Göta älvutredningen, GÄU. Statens geotekniska institut, SGI. Varia 624:1, Linköping.
  50. Fredén, C. (1984) Beskrivning till jordartskartan Vänersborg SO. Sveriges Geologiska Undersökning, Serie Ae nr 48.
  51. Fredén, C. (1986) Beskrivning till jordartskartan Göteborg NO. Sveriges Geologiska Undersökning, Serie Ae nr 40.
  52. Klingberg, F. (2011) Bottenförhållanden i Göta älv. Statens geotekniska institut, SGI. Göta älvutredningen, GÄU. Delrapport 6. Även publicerad som SGU-rapport 2010:7, Linköping.
  53. Klingberg, F., Påsse, T. and Levander, J. (2006) Bottenförhållanden och geologisk utveckling i Göta älv. Sveriges Geologiska Undersökning, SGU. Rapport K43, Uppsala.
  54. Hedvall, T. (2010) Bearbetning av Terrängmodell över Göta och Nordre Älv. Vattenfall Power Consultant AB.
  55. Marin, M. (2011) Ytgeologisk undersökning med backscatter-analys för Göta älv och Nordre älv. arin Miljö- analys AB, U304-0909. Statens geotekniska institut, SGI. Göta älvutredningen, GÄU. Delrapport 5 and Delrapport 34, Linköping.
  56. Persson, H., Bengtsson, P.E., Lundström, K. and Karlsson, P. (2011) Bedömning av grundvattenförhållanden för slänter längs Göta älv. Allmän vägledning. Statens geotekniska institut, SGI. Göta älvutredningen, GÄU. Delrapport 7, Linköping.
  57. Millet, D. (2011) River erosion, landslides and slope development in Göta river—A study based on athymetric data and general limit equilibrium slope stability analysis. Chalmers University of Technology. department of Civil and Environmental Engineering. GeoEngineering. Geology and Geotechnical Engineering. Master thesis 2011:131, Göteborg.
  58. Berggren, B., Alén, C., Bengtsson, P.E. and Falemo, S. (2011) Metodbeskrivning sannolikhet för skred: Kvantitativ beräkningsmodell (Methodology for assessment of landslides probability: Quantitative model). Statens Geotekniska Institute, SGI. Report28, Linköping.
  59. Janbu, N. (1954) Stability analysis of slopes with dimensionless parameters. Doctoral Thesis. Harvard University, Cambridge.
  60. [61] Alén, C. (1998) On probability in geotechnic. Random calculation models exemplified on slope stability analysis and ground-superstructure interaction. Doctoral Thesis, Department of Geotechnical Engineering, Chalmers University of Technology, Göteborg.
  61. [62] Alén, C., Bengtsson, P.E., Berggren, B., Johansson, L. and Johansson, Å. (2000) Skredriskanalys i Göta älvdalenMetodbeskrivning (Landslide risk analysis in Göta rivermethodology). Statens Geotekniska Institut, SGI. Report 58, Linköping.
  62. [63] Möller, B. and Bergdahl, U. (1982) Estimation of the sensitivity of soft clays from static and weight sounding tests. Proceedings of the European Symposium on Penetration Testing, 2, ESPOT2, Amsterdam, 1, 291-295.
  63. [64] Rankka, K., Andersson-Sköld, Y., Hulten, C., Larsson, R., Leroux, V. and Dahlin, T. (2004) Quick clay in Sweden. Swedish Geotechnical Institute, Report 65, Linköping.
  64. [65] SCB (2003) Time for daily activities and living conditions (Tid för vardagsliv, Levnadsförhållanden, in Swedish). Statistiska Centralbyrån SCB 99.
  65. [66] MSB (2011) National database of natural hazards, Swedish Civil Contingencies Agency. http:/ndb.msb.se
  66. [67] NGU och NVE (2010) “Skrednett”. Norges vassdragsog energidirektorat. www.skrednett.no
  67. [68] Berggren, B., Fallsvik, J., Hintze, S. and Stille, H. (1991) Lerslänters stabilitetsförhållanden. Riskvärdering och eslutsteori. Förslag till metod för riskvärdering. Statens Geotekniska Institut, SGI. Varia 333, Linköping.
  68. [69] Blied, L. and Persson, H. (2011) Bedömd förändring av maximala grundvattennivåer i Göta älvdalen till följd av förändrat klimat. Statens Geotekniska Institut. SGI. Göta Älvutredningen, GÄU. Delrapport 9, Linköping.
  69. [70] Kjellgren, S. (2013) Exploring local risk managers use of flood hazard maps for risk communication purposes in Baden-Württemberg. Natural Hazards & Earth System Sciences, 13, 1857-1872.

NOTES

1Sampers is a Swedish National Travel Demand Forecasting Tool (applied by Röcklinger as described in Andersson-Sköld, 2011).

2Road Analyses Effects (Effekter vid väganalyser).

3http://www.lansstyrelsen.se/vastragotaland/SiteCollectionDocuments/Sv/manniska-och-samhalle/krisberedskap/smhi-rapport201085.pdf.