Nat Hazards (2011) 56:465480 DOI 10.

1007/s11069-010-9703-6 ORIGINAL PAPER

Avalanche risk assessment for mountain roads: a case study from Iceland
tter Hannes Kleindienst Maria Wastl Johann Sto

Received: 24 November 2008 / Accepted: 23 December 2010 / Published online: 6 January 2011 Springer Science+Business Media B.V. 2011

Abstract This paper presents an assessment of the avalanche hazard potential and the resulting risks on mountain roads for a 38.7-km-long section of road no 76 (Siglufjararvegur) in northern Iceland following a regional scale approach developed in the Alps. The assessment of the individual avalanche death risk proved applicable to distinguish areas of avalanche hazard with a risk above the accepted level, which should be given priority in following detailed investigations and the planning of possible protective measures, from road sections where the avalanche death risk is low and accepted according to international practice. The cumulative individual and collective avalanche death risks in the investigated road section provide a comparable measure for assessing the avalanche hazard both within the Icelandic public road network and on an international scale. The case study on road no 76 in northern Iceland shows that a standardised regional scale riskbased approach is practical to determine, analyse and assess the avalanche hazard situation on mountain roads in Iceland and guarantees comprehensible, reproducible and comparable results as a basis for a sustainable planning of measures. Keywords Risk assessment Avalanche hazard Mountain roads Iceland

M. Wastl (&) Department of Geography, Institute for Integrated Natural Sciences, Koblenz University, tsstrae 1, 56070 Koblenz, Germany Universita e-mail: wastl@uni-koblenz.de URL: http://www.uni-koblenz-landau.de/koblenz/fb3/in/geographie tter J. Sto Institute of Geography, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria H. Kleindienst r angewandte Geoinformatik mbH, Technikerstrae 21a, GRID-IT Gesellschaft fu 6020 Innsbruck, Austria

123

466

Nat Hazards (2011) 56:465480

1 Background and aim of the investigation While the management of avalanche hazards in settlements follows operational stand hannesson 2004; Jo nasson ardised procedures in Iceland (see e.g. Arnalds et al. 2004; Jo nsson 2002; http://www.vedur.is/ofanod/log/), there are no comparable et al. 1999; Jo approaches for a systematic survey and assessment of the avalanche hazard situation and a sustainable planning of measures for roads. This investigation presents a regional scale assessment of the avalanche hazard potential and its related risks on mountain roads for a case study in northern Iceland. The total length of the public road network in Iceland is ca. 13,000 km (http://www. vegagerdin.is/vefur2.nsf/Files/VegskraLysing/$le/Vegaskr%C3%A1_lei%C3%B0arl%C3% BDsing_01-03-2010.pdf), mostly low-volume roads outside built-up areas. About 10,500 km of these roads are open all year. Substantial parts of the Icelandic public road network e.g. in central northern Iceland, north-western and eastern Iceland lie in alpine mountain areas and are affected by avalanches. Though the resulting road maintenance costs are considerable, there is no general overview of the avalanche hazard situation up to now. (Fig. 1) The aim of the investigation is to describe and assess the generalised avalanche hazard situation and its related risks on roads outside built-up areas following a regional scale approach. The approach is calibrated by a detailed investigation into the 38.7-km-long section of rur to Straumnes in northern Iceland (Fig. 2). road no 76 (Siglufjararvegur) from Siglufjo The results are documented in maps indicating the avalanche hazard potential for specic road sections, a report with the risk assessment and recommendations and an information system, all together providing a basis for detailed planning. To guarantee that investigations following the developed approach are feasible and affordable for all rural roads in Iceland and reduce time-consuming and expensive eldwork to a minimum, the investigation needs to be mainly based on already existing data most of which are available at the Icelandic Road Administration (Vegagerin), the Iceslands) and the National Land Survey of Iceland landic Meteorological Ofce (Veurstofa I (Landmlingar Islands).

Fig. 1 Road no 76 in northern Iceland

123

Nat Hazards (2011) 56:465480

467

rur to Straumnes. The orange line marks the Fig. 2 Investigated section of road no 76 from Siglufjo boundary of the catchments of the streams crossing the road, the blue dots indicate the meteorological stations used in this investigation

This paper describes the practical application of a regional scale risk-based avalanche hazard assessment for mountain roads in Iceland meeting these preconditions.

2 Investigated road section Road no 76 is classied as a highway and tarred. The mean trafc densities along the investigated section of the road are between ca. 250 and 370 cars per day in the annual average, with ca. 400540 cars per day during the summer (June to September) and ca. 150270 cars per day during the winter months (December to March), based on the census from 2008 (http://www.vegagerdin.is/vefur2.nsf/Files/umferd_thjod_2008_skyrsla/$le/ Umfer%C3%B0_%C3%A1_%C3%BEj%C3%B3%C3%B0vegum_2008.pdf). Figure 2 shows the investigated section of road no 76. For the assessment of the avalanche hazard situation, it was divided into six subsections (road sections 911941 in the map).

123

468

Nat Hazards (2011) 56:465480

3 Avalanche hazard assessment 3.1 Method Due to the specications stated in the objectives (see Sect. 1) and the requirements of the regional scale and the resulting detail in the determination of areas affected by avalanche hazards, an approach was adopted which had been developed in the Eastern Alps in recent years (see http://bfw.ac.at/iym/pdf/ziegner.pdf; http://www.tirol.gv.at/leadmin/www.tirol. gv.at/themen/umwelt/wald/schutzwald/downloads/endber-kurzfass-06-04-og.pdf). In addition to the natural hazard zone maps on the planning level with scales C1:10,000, a preceding regional scale overview level with scales B1:25,000 has been introduced in Austria, Germany and Italy. This approach aims at increasing the efciency and effectiveness in natural hazard management by maximising the use of existing data (maps, reports, scientic investigations, etc.) and reducing time-consuming and expensive eldwork to a minimum. Models which are specially adapted to the requirements of a regional scale are applied to provide comprehensible, reproducible and comparable results, which help to assign priorities in the planning of measures. 3.2 Data bases To guarantee that the developed approach can be applied to all classied roads in Iceland, the investigation was mainly based on already existing data most of which are available at the ofces of the Icelandic Road Administration, i.e. (1) (2) digital colour orthophotos with a resolution of 0.5 m and digital 5 m contour lines

for the areas around the road. The mosaic of the digital colour orthophotos provides the background for 3D-views of the investigated road sections and the detailed maps of the results (see Sect. 3.3). the Icelandic Meteorological Ofce, i.e. meteorological data of stations in or close to the investigation area: Sauanesviti (66110 N, 18570 W, 30 m a.s.l.), Skeisfoss (66000 N, 19010 W, 84 m a.s.l.) and Siglunes (66120 N, 18510 W, 8 m a.s.l.) (see Fig. 2). the National Land Survey of Iceland, i.e. (1) (2) sland 1:50,000 and topographic maps Iceland-I black and white aerial photographs.

Stereo pairs of these aerial photographs provided the basis for mapping potential starting areas of avalanches above the investigated section of road no 76 before going into the eld (see Sect. 3.3). Orthophotos of a resolution of 1.0 m were produced from the black and white aerial photographs by means of the software PCI OrthoEngine. For this investigation, a digital terrain model (DTM) was produced using ARC-GIS. It is based on the digitised contour lines (interval 20 m) and geodetic points of the 1:50,000 maps for the whole investigation area and the digital 5-m contour lines provided by the Icelandic Road Administration for parts of the area along the investigated section of road no 76. The elevation data of the 5-m contours were used where available. This improved

123

Nat Hazards (2011) 56:465480

469

DTM provided the basis for modelling the investigated processes, which guarantees a high quality of the modelling results for the areas around the road. In addition, these data studies on the general geological background (Gumundsson et al. 2001; Haiason 1982) as well as previously made reports on the investigated road GVST, BAH Ra gjo f 1999; Smundsson et al. 2004) were section (Jarfristofa A available for this assessment. The latter, however, concentrate on an area of active sagging and do not provide information on the avalanche hazard situation. 3.3 Applied model Objective of the modelling part of the investigation was the determination and outline of rur to areas affected by avalanche processes for the section of road no 76 from Siglufjo Straumnes. Potential starting areas of avalanches were mapped based on stereo pairs of the aerial photographs (see Sect. 3.2) for the areas above the road. These maps were checked and completed in the eld. Areas of active avalanche processes could be identied by damage to the vegetation, missing sods or eroded patches of soil. In some cases, remains of snow or debris left by avalanches could be found. Boulders deposited by the avalanches were mapped as indicators for the extents of the accumulation areas. Additional information on the run-out distance was provided by ruins of farm buildings which had been destroyed by an avalanche (Engidalur). The ground-truthed maps were digitised and transferred to a GIS dataset. The starting areas were checked and partly expanded by means of a disposition model using functionalities of geographical information systems. On the regional scale, the most important criterion for the determination of the starting areas is the slope of the terrain. Potential starting lines of avalanches are dened as the upper boundaries of areas with slopes between 28 and 50. The resulting dataset of potential avalanche starting areas was the basis for modelling the accumulation areas using the model GRID-aval. The avalanche model GRID-aval (Grid-based Trajectory Avalanche Model) is a twodimensional model to calculate the run-out distance of ow avalanches on a regional scale tter et al. 2006). It is based on the approach of Lied and Bakkehi (1980) and (see Sto determines the maximum run-out distance in a topographic-statistical way. The model describes the run-out distance of an avalanche by means of an angle a which is a function of the angle between the starting point and the point at which the avalanche track reaches an angle of 10. The model calculates the run-out distance of an avalanche starting from a given line by means of an estimated slope with the angle a. This estimated slope describes the loss of energy of the avalanche during the ow process. Thus, the difference between the real slope of the terrain surface and the estimated slope gives the corresponding kinetic energy, which, in a following step, can be used as input data for the modelling of the ow paths. The angle a is determined on the basis of a statistical analysis of known avalanche events. For this assessment, a was derived from the statistical analysis of 45 avalanche events hannesson (1998).This investigation gives a mean value for a of 23.62 in Iceland by Jo with a standard deviation of 3.19 (a = 23.62, r = 3.19, n = 45). The minimum value rur area (n = 7) is 21. On this basis, an estimated slope of 23.62 for a in the Siglufjo (mean) and a lower limit of 20.43 (mean minus one standard deviation) were used in the calculations.

123

470

Nat Hazards (2011) 56:465480

The avalanche track is determined by means of a vector-orientated model on the basis of orientation and slope data from the digital terrain model. The inuence of topography on the calculated avalanche tracks is determined by a weighting factor (topoweight). A topoweight value of 1 stands for a motion close to that of a ball while a very high value represents the motion characteristics of water. A further model parameter (widening) denes the lateral extension of the avalanche from the calculated tracks. In model calculations made so far, topoweight values between 5 and 7 and a widening of 0.005 have given good results. The output of this modelling is a sequence of points in three-dimensional coordinates. The modelled avalanche accumulation areas were printed out against the background of the topographic maps and the colour orthophotos (see Sect. 3.2) and compared with the extents of the avalanche accumulation areas mapped in the eld. These checked and corrected data provided the basis for assessing the avalanche hazard for the investigated road section. Based on the ndings in the eld and the modelling results, two avalanche hazard levels are distinguished: (1) Avalanche hazard level 1 is based on an estimated slope of 23.62, which is the hannesson statistical mean of the analysis of 45 avalanche events in Iceland by Jo (1998).

Areas of avalanche hazard level 1 can be reached by avalanches under unfavourable conditions. (2) Avalanche hazard level 2 is based on an estimated slope of 20.43, which is the mean hannesson (1998), and indicates minus one standard deviation in the analysis by Jo areas that can be reached by avalanches under extremely unfavourable conditions (worst case assumption).

There are, however, no records of avalanches for the investigated road which could provide empirical evidence for the assumptions on these avalanche hazard levels. Figure 3 shows an example of the modelled avalanche tracks and the extent of the accumulation areas, which are the outer boundaries of all cells reached by the avalanche tracks, for avalanche hazard levels 1 and 2, respectively. While the validation of run-out distances is not possible due to the limitations of the available data, the avalanche hazard levels reect the topographic situation of the affected road sections, where the slopes with the avalanche tracks extend down to the coastline (see Fig. 1). Thus, all modelled avalanche tracks that reach or cross the road cause an avalanche hazard in these areas irrespective of their exact run-out distances. 3.4 Results rur and Straumnes, there are 18 In the investigated part of road no 76 between Siglufjo areas of avalanche hazard level 1 (see Table 1) and 25 areas of avalanche hazard level 2 (see Table 2). rur and km 13.32, i.e. in The areas of avalanche hazard level 1 lie between Siglufjo road sections 911, 912 and 921 (see Fig. 2). Altogether, 6.81 km of these northern three sections can be reached by avalanches under unfavourable conditions (Figs. 4, 5, 6). Especially, road sections 911 and 912 are almost entirely areas of avalanche hazard level 1 with 14 avalanche tracks, some of which very broad, crossing this part of the road.

123

Nat Hazards (2011) 56:465480

471

Fig. 3 Modelled avalanche tracks and accumulation areas for avalanche hazard levels 1 (dark blue) and 2 tter et al. 2006) (light blue) (Sto

Table 1 Areas of avalanche hazard level 1 on road no 76 from rur to Straumnes Siglufjo

No

Road km from rur Siglufjo 00.39 0.881.08 1.391.46 1.502.00 2.082.17 3.194.43 4.795.03 5.095.29 5.395.66 5.715.73 5.887.32 7.457.48 7.587.99 8.539.20 9.549.82 11.0911.41 12.2212.39 13.0513.32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 The total length of the areas of avalanche hazard level 1 is 6.81 km 17 18

123

472 Table 2 Areas of avalanche hazard level 2 on road no 76 from rur to Straumnes Siglufjo

Nat Hazards (2011) 56:465480

No

Road km from rur Siglufjo 00.42 0.881.08 1.382.17 3.174.54 4.695.06 5.095.30 5.395.66 5.715.73 5.867.38 7.438.03 8.529.23 9.509.82 11.0811.44 11.9612.15 12.2112.46 13.0313.35 13.4513.52 13.6113.63 13.7713.83 13.9514.05 14.1014.18 14.2114.24 31.9932.66 35.7435.76 35.8335.85

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 The total length of the areas of avalanche hazard level 2 is 8.99 km 24 25

Fig. 4 Areas of avalanche hazard for avalanche hazard levels 1 (dark blue) and 2 (light blue) in road section 911: km rur to the 03.2Siglufjo northernmost point of road no 76

The areas of avalanche hazard level 1 in section 921 are limited to four comparatively short parts of the road. West of km 13.32 the slopes along road no 76 are less steep and less high. There are steep and high slopes in road section 941 (see Fig. 2). These are, however, so far away from the road that the modelled accumulation areas for avalanche hazard level 1 of the four potential avalanches in this section do not reach the road (see Fig. 7).

123

Nat Hazards (2011) 56:465480

473

Fig. 5 Areas of avalanche hazard for avalanche hazard levels 1 (dark blue) and 2 (light blue) in road k section 912: km 3.29.5northernmost point of road no 76 to Skrinav

Fig. 6 Areas of avalanche hazard for avalanche hazard levels 1 (dark blue) and 2 (light blue) in road k to south of Hraun section 921: km 9.518.7Skrinav

Fig. 7 Areas of avalanche hazard (avalanche hazard level 2) in road section 941: km 31.138.7pipe 510 to road to Straumnes

The modelled accumulation areas of avalanches for avalanche hazard level 2 are generally larger than those of avalanche hazard level 1. The areas of avalanche hazard level 2 thus usually form a band around the areas of avalanche hazard level 1 (see Fig. 3). For road sections 911, 912 and 921, this means that 8.28 km of the road can be reached by avalanches under extremely unfavourable conditions (worst case assumption). While in road sections 911 and 912 with their extensive areas of avalanche hazard level 1, there are only small further areas of avalanche hazard level 2, ten additional avalanches reach the road in section 921 under the conditions of avalanche hazard level 2. The total extent of the areas of avalanche hazard level 2 in this road section remains however limited. Three further areas of avalanche hazard level 2 lie in road section 941 (Fig. 7).

4 Risk assessment The assessment of the avalanche risk for the investigated section of road no 76 is based on investigations by Wilhelm (1997, 1998, 1999) in the Alps. This approach has been applied

123

474

Nat Hazards (2011) 56:465480

e.g. to high alpine pass roads in Switzerland (Margreth et al. 2003), to mountain roads in Italy (Zischg et al. 2004, 2005) and Austria (Guer 2007; Huttenlau 2004) and to roads in rur and Norway (Kristensen et al. 2003). For the study on road no 76 between Siglufjo Straumnes, it was adapted to comply with the data availability of the regional scale. According to this approach, there is a distinction between (1) the individual risk and the collective risk for the society for each area of avalanche hazard and (2) the cumulative individual and collective risks for the investigated road section. The individual death risk (1.1), collective death risk (1.2), cumulative individual death risk (2.1) and cumulative collective death risk (2.2) can be determined by the following equations: rind rcol rind
i i

1 gi zi k Ti vi 24h

1:1 1:2 2:1

1 gi WDT bk Ti vi 24h rind

i

sum

n X i 1

n zi X gi k 24h i1 Ti vi

rcol with rind_i rcol_i rind_sum rcol_sum Ti gi zi vi k WDT i = 1, 2,,n

sum

n X i 1

rcol

n WDT b X gi k 24h T vi i i 1

2:2

for individual death risk for avalanche hazard area i [1/years], for collective death risk for avalanche hazard area i [deaths/year], for cumulative individual death risk for n areas of avalanche hazard [1/ years], for cumulative collective death risk for n areas of avalanche hazard [deaths/year], for mean return period of the avalanche [years], for width of the area of avalanche hazard [km], for number of passages per day, for mean velocity in the investigated road section [km/h], for mean death rate in cars involved in avalanches, for average daily winter trafc [cars], for degree of occupancy [persons/car] and areas of avalanche hazard

As the input data required for these equations are only partly available for the investigated section of road no 76, the following assumptions are made for the risk assessment. Based on the census from 2008 (http://www.vegagerdin.is/vefur2.nsf/Files/umferd_ thjod_2008_skyrsla/$le/Umfer%C3%B0_%C3%A1_%C3%BEj%C3%B3%C3%B0vegum_ 2008.pdf), an average trafc density (WDT) of 270 cars is used for road no 76 between rur and the tunnel (see Fig. 4) during the winter months and a WDT of 150 cars Siglufjo for the section of the road from the tunnel to Straumnes. The number of passages (zi) is assumed to be one for the areas of avalanche hazard from the northern end of the tunnel to rur and the southern end of the tunnel as part of the Straumnes, and 1.75 between Siglufjo rur. There are no cars moving in this section of the road reect local trafc from Siglufjo

123

Nat Hazards (2011) 56:465480

475

data on the mean velocity (vi) and degree of occupancy () of the cars. For this assessment, the former is estimated to be 60 km/h and the latter two persons/car. Data on the death rate in cars hit by avalanches (k) are not available in Iceland. According to the statistical analyses of avalanche accidents on roads in Switzerland by Wilhelm (1999), 18% of the people who are hit by avalanches in their cars die. For more remote areas with unfavourable topographic characteristics in Norway, Kristensen et al. (2003) assume a death rate of 40% for avalanche accidents due to longer rescue times. On s from where rur and Hofso road no 76, the distance between the settlements of Siglufjo rescue operations could start is about 60 km. Therefore, long rescue times should not have a particular effect on the death rate in the investigated section of this road, and a death rate of 18% as determined by Wilhelm (1999) is used in the assessment of the avalanche risk. The length of the road sections affected by avalanches (gi) is derived from the modelled areas of avalanche hazard (see Sect. 3.4). The most difcult assumption concerns the mean return period (Ti) of the avalanches as avalanche inventories are not available. Against this background, three groups of return periods are distinguished on the basis of the reconstructed areas of avalanche hazard (see Sect. 3.3): (1) Avalanches with a long return period (Ti = 100)

These are avalanches which reach the road only under extremely unfavourable conditions as dened for avalanche hazard level 2 (see Sect. 3.3). (2) Avalanches with a short return period (Ti = 10)

These are avalanches for which the reconstructed areas of avalanche hazard level 1 (see Sect. 3.3) reach far beyond the road and which thus come down to the road even in the case of relatively small events. (3) Avalanches with a medium return period (Ti = 30)

These are avalanches for which the reconstructed areas of avalanche hazard level 1 (see Sect. 3.3) reach only little beyond the road and which thus only seldom come down to the road. The assessment of the individual death risk rind_i [1/years] for each area of avalanche rur to Straumnes identies three areas of avahazard on road no 76 from Siglufjo lanche hazard with an rind_i greater than 1 9 10-5 (orange colouring in Table 3). These sections should be given priority for following detailed investigations and the planning of possible protective measures. Twelve areas of avalanche hazard have an rind_i between 1 9 10-5 and 1 9 10-6 (yellow colouring in Table 3), while in the rest of the areas of avalanche hazard the rind_i is less than 1 9 10-6. The results in Table 3 show that both the individual death risk rind_i [1/years] and the collective death risk rcol_i [deaths/year] are very sensitive to the lengths of the road sections in the areas of avalanche hazard. In the cases with an rind_i greater than 1 9 10-6 the modelled areas of avalanche hazard are often very wide and thus affect long sections of the road. If it is assumed, however, that not all avalanche events reach the maximum possible extent the result looks different. For an event reaching half of the maximum possible modelled width the rind_i remains below 1 9 10-5 everywhere in the investigated road section and exceeds 1 9 10-6 in eleven areas of avalanche hazard altogether. The cumulative individual death risk rind_sum [1/years] for all areas of avalanche hazard rur and Straumnes, under the assumptions stated above on road no 76 between Siglufjo

123

476

Nat Hazards (2011) 56:465480

Table 3 Individual death risk rind_i [1/years], collective death risk rcol_i [deaths/year], cumulative individual death risk rind_sum [1/years] and cumulative collective death risk rcol_sum [deaths/year] for road no 76 rur to Straumnes with gi for width of the area of avalanche hazard [km], Ti for mean return from Siglufjo period of the avalanche [years], zi for number of passages per day, vi for mean velocity in the investigated road section [km/h], k for mean death rate in cars involved in avalanches, WDT for average daily winter trafc [cars] and for degree of occupancy [persons/car]
Road km 0.00-0.39 0.39-0.42 0.88-1.08 1.38-1.39 1.39-1.46 1.46-1.50 1.50-2.00 2.00-2.08 2.08-2.17 3.17-3.19 3.19-4.43 4.43-4.54 4.69-4.79 4.79-5.03 5.03-5.06 5.09-5.29 5.29-5.30 5.39-5.66 5.71-5.73 5.86-5.88 5.88-7.32 7.32-7.38 7.43-7.45 7.45-7.48 7.48.7.58 7.58-7.99 7.99-8.03 8.52-8.53 8.53-9.20 9.20-9.23 9.50-9.54 9.54-9.82 11.08-11.09 11.09-11.41 11.41-11.44 11.96-12.15 12.21-12.22 12.22-12.39 12.39-12.46 13.03-13.05 13.05-13.32 13.32-13.35 13.45-13.52 13.61-13.63 13.77-13.83 13.95-14.05 14.10-14.18 14.21-14.24 31.99-32.66 35.74-35.76 35.83-35.85 rind_sum [1/years]/ rcol_sum [deaths/year] gi [km] 0.39 0.03 0.20 0.01 0.07 0.04 0.50 0.08 0.09 0.02 1.24 0.11 0.10 0.24 0.03 0.20 0.01 0.27 0.02 0.02 1.44 0.06 0.02 0.03 0.10 0.41 0.04 0.01 0.67 0.03 0.04 0.28 0.01 0.32 0.03 0.19 0.01 0.17 0.07 0.02 0.27 0.03 0.07 0.02 0.06 0.10 0.08 0.03 0.67 0.02 0.02 Ti [years] 10 100 10 100 10 100 10 100 10 100 10 100 100 10 100 10 100 10 10 100 10 100 100 30 100 10 100 100 10 100 100 10 100 30 100 100 100 30 100 100 30 100 100 100 100 100 100 100 100 100 100 zi 1.75 1.75 1.75 1.75 1.75 1.75 1.75 1.75 1.75 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 vi [km/h] 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 60 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 WDT [cars] 270 270 270 270 270 270 270 270 270 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 150 [persons/car] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 rind_i [1/years] 0,00000853 0,00000007 0,00000438 0,00000002 0,00000153 0,00000009 0,00001094 0,00000018 0,00000197 0,00000003 0,00001550 0,00000014 0,00000013 0,00000300 0,00000004 0,00000250 0,00000001 0,00000338 0,00000025 0,00000003 0,00001800 0,00000008 0,00000003 0,00000013 0,00000013 0,00000513 0,00000005 0,00000001 0,00000838 0,00000004 0,00000005 0,00000350 0,00000001 0,00000133 0,00000004 0,00000024 0,00000001 0,00000071 0,00000009 0,00000003 0,00000113 0,00000004 0,00000009 0,00000003 0,00000008 0,00000013 0,00000010 0,00000004 0,00000084 0,00000003 0,00000003 0,00009314 rcol_i [deaths/year] 0,00263250 0,00002025 0,00135000 0,00000675 0,00047250 0,00002700 0,00337500 0,00005400 0,00060750 0,00000750 0,00465000 0,00004125 0,00003750 0,00090000 0,00001125 0,00075000 0,00000375 0,00101250 0,00007500 0,00000750 0,00540000 0,00002250 0,00000750 0,00003750 0,00003750 0,00153750 0,00001500 0,00000375 0,00251250 0,00001125 0,00001500 0,00105000 0,00000375 0,00040000 0,00001125 0,00007125 0,00000375 0,00021250 0,00002625 0,00000750 0,00033750 0,00001125 0,00002625 0,00000750 0,00002250 0,00003750 0,00003000 0,00001125 0,00025125 0,00000750 0,00000750 0,02817800

Areas of avalanche hazard with an rind_i greater than 1 9 10-5 are marked in orange, those with an rind_i between 1 9 10-5 and 1 9 10-6 in yellow

and for the case that no measures of avalanche prevention or control or temporary closing of the road are taken, is 9.314 9 10-5. The cumulative collective death risk rcol_sum [deaths/year] for the investigated road section is 0.0282, or one avalanche death in 36 years.

123

Nat Hazards (2011) 56:465480

477

This cumulative individual death risk is one order of magnitude less than the value ela Pass in the Swiss Alps for the state without safety determined for the road over the Flu measures by Margreth et al. (2003) (see the comparison of the cumulative individual death rur and risk and the cumulative collective death risk on road no 76 between Siglufjo Straumnes with those on roads in the Alps in Table 4). It is also comparable to or less than the cumulative individual death risks found e.g. by Huttenlau (2004) and Guer (2007) for roads in inner Stubai and inner Oetz Valley in Austria, which include the risk reduction by permanent measures, or by Zischg et al. (2005) for the Sulden road in Italy. The cumulative collective death risk for the investigated section of road 76 is very low compared to mountain roads in the Alps.

5 Conclusions According to Wilhelm (1999), the accepted individual avalanche death risk rind_i [1/years] for the user of a public road is \1 9 10-5. Against this background, the assessment of the rur to individual death risk for each area of avalanche hazard on road no 76 from Siglufjo Straumnes can be used to distinguish areas of avalanche hazard with a risk level above the critical value, which should be given priority for following detailed investigations and the planning of possible protective measures, from road sections where the avalanche death risk is low and accepted based on international practice (see Table 3). The calculated risks for the investigated section of road no 76 can further be reduced by measures like temporary closing of parts of the road. This requires, however, a systematic monitoring of the development of the avalanche hazard situation along the road. The cumulative individual and collective avalanche death risks in the investigated road section, on the other hand, provide a reproducible and comparable measure for assessing the avalanche hazard situation and the related risks both within the Icelandic public road network and on an international scale (see Table 4). In this context, it is interesting to note that the cumulative individual avalanche death risk rind_sum [1/year] determined for road no rur and Straumnes of 9.314 9 10-5 is equal to the average trafc 76 between Siglufjo mortality rate in Iceland in the last 10 years, i.e. 9 9 10-5 (http://www.us.is/id/4219). The investigation into the avalanche hazard situation and the related risks on road no 76 rur and Straumnes in northern Iceland shows that a standardised regional between Siglufjo scale risk-based approach is practical to determine, analyse and assess the avalanche hazard situation on mountain roads in Iceland and guarantees comprehensible, reproducible and comparable results, which help to assign priorities in following detailed investigations and the planning of measures. For the natural hazard management, the results of this assessment need to be combined with road data (e.g. current and expected trafc volume for various timescales, road clearing and maintenance costs for problematic road sections). This allows cost benet analyses as a basis for decisions on protective measures. To meet these requirements, an tter et al. 2006). Internet Road Information System was developed (see Sto In the last decades, Iceland used a lot of money to improve the public road network and the accessibility of remote communities during the winter (see http://www. vegagerdin.is/upplysingar-og-utgafa/aaetlanir/vegaaetlun/) often by building tunnels in road sections with a high avalanche hazard potential (http://www.vegagerdin.is/vegakerd/ jardgong/jardgvegakerf/; http://www.vegagerdin.is/storframkvaemdir/hedinsfjardargong/; http://www.vegagerdin.is/storframkvaemdir/oshlidargong/). This was part of a policy to provide an equally high level of infrastructure in all parts of the country and thus work

123

478

Nat Hazards (2011) 56:465480

Table 4 Comparison of the cumulative individual death risk rind_sum [1/years] and the cumulative col rur and Straumnes with values lective death risk rcol_sum [deaths/year] on road no 76 between Siglufjo determined for roads in the Alps Investigated road section rind_sum [1/years] 9.314 9 10-5 rcol_sum [deaths/year] 0.0282

rur and Road no 76 between Siglufjo Straumnes, northern Iceland (38.7 km, 25 avalanche tracks, 9 km areas of avalanche hazard), without safety measures (this paper) ela Pass, Switzerland Road over the Flu (19.3 km, 47 avalanche tracks, endangered road length 10.1 km), without safety measures (Margreth et al. 2003) Road from Ranalt to the terminal of the Stubaier Gletscherbahn, inner Stubai Valley, Tyrol, Austria (7.24 km, 16 avalanche tracks), with permanent measures (Huttenlau 2004) Road from Zwieselstein to Obergurgl, inner Oetz Valley, Tyrol, Austria (10 km, 10 avalanche tracks), with permanent measures (Guer 2007) Road from Zwieselstein to Vent, inner Oetz Valley, Tyrol, Austria (13 km, 21 avalanche tracks), with permanent measures (Guer 2007) Sulden road between Prad and Sulden, Ortles Alps, Southern Tyrol, Italy (summer road, 20 km, 22 avalanche tracks), without safety measures (Zischg et al. 2005) Sulden road between Prad and Sulden, Ortles Alps, Southern Tyrol, Italy (winter road, 20 km, 17 avalanche tracks), without safety measures (Zischg et al. 2005)

8.7 9 10-4

0.70

3.65.4 9 10-5

0.19860.4252

1.221 9 10-4

1.100

4.523 9 10-4

2.378

1.16 9 10-4

0.113

0.0798

against the rural exodus. As a consequence of the current nancial crisis in Iceland, public spending on the road network has been cut back considerably and will be so for the foreseeable future. Against this background, an approach as it is presented in this paper can become even more important both as a basis for a sustainable planning of measures and for the communication of risk to the public.
Acknowledgments This paper is based on a study for the Icelandic Road Administration made at the nuho nnun alpsCentre for natural hazard and risk management GmbH in Innsbruck in cooperation with L k. We thank the Icelandic Road Administration and the Icelandic Meteoconsulting engineers in Reykjav r G. Pe tursson at the Icelandic Institute of Natural rological Ofce for the provision of data and Dr. Halldo History in Akureyri for useful information on the investigated road section. We also thank the reviewers for their constructive comments on the paper.

References
nasson K, Sigursson S (2004) Avalanche hazard zoning in Iceland based on individual risk. Arnalds , Jo Ann Glaciol 38:285290

123

Nat Hazards (2011) 56:465480

479

hannesson T (1998) Icelandic avalanche runout models compared with topographical models used in other Jo countries. In: Hestnes E (ed) 25 years of snow avalanche research, vol 203. NGI Publ, Oslo, pp 4352 R hannesson T (2004) Greining a t ni snjo o a notkun hu sa a vetrarlagi. U rkun a Jo sambandi vi takmo slands, Reykjav J200407, Veurstofa I k To R99001U R01, Veurstofa nasson K, Sigursson S, Arnalds (1999) Estimation of avalanche risk. VI Jo slands, Reykjav k I nsson T (2002) Httumat og hlutverk Veurstofunnar si httumatsramma Aljo aveurfristofJo ljo U R11, Veurstofa I slands, Reykjav k nunarinnar. G02021 VI Kristensen K, Harbitz CB, Harbitz A (2003) Road trafc and avalanchesmethods for risk evaluation and risk management. Surv Geophys 24:603616 Lied K, Bakkehi S (1980) Empirical calculations of snow-avalanche run-out distance based on topographical parameters. J Glaciol 26:165177 Margreth S, Stoffel L, Wilhelm C (2003) Winter opening of high alpine pass roadsanalysis and case studies from the Swiss Alps. Cold Reg Sci Tech 37:467482 rsson H, Kleindienst H (2006) GIS-based regional scale tter J, Wastl M, Scho berl F, Lippert J, Sigo Sto rur to assessment of the natural hazard potential for roadscase study road no 76 from Siglufjo Straumnes, Iceland. In: Strobl J, Roth C (eds) GIS und Sicherheitsmanagement. Wichmann, Heidelberg, pp 97106 Wilhelm C (1997) Wirtschaftlichkeit im Lawinenschutz. Methodik und Erhebungen zur Beurteilung von konomischer Bewertung. Mitt Eidgeno ss Schutzmassnahmen mittels quantitativer Risikoanalyse und o Inst Schnee- und Lawinenforsch 54, Davos Wilhelm C (1998) Quantitative risk analysis for evaluation of avalanche protection projects. NGI Publ 203:288293 Oslo Wilhelm C (1999) Kosten-Wirksamkeit von Lawinenschutzmassnahmen an Verkehrsachsen. Vorgehen, r Umwelt, Wald und Landschaft (ed) Beispiele und Grundlagen der Projektevaluation. Bundesamt fu Vollzug Umwelt, Praxishilfe, Bern tter J (2004) Uncertainties and fuzziness in analysing risk related to natural hazards Zischg A, Fuchs S, Sto a case study in the Ortles Alps, South Tyrol, Italy. In: Brebbia CA (ed) Risk Analysis IV, Management Information Systems 9, Southampton, pp 523532 tter J (2005) Temporal variability of damage potential on roads as a Zischg A, Fuchs S, Keiler M, Sto conceptual contribution towards a short-term avalanche risk simulation. Nat Haz Earth Sys Sci 5:235242

Online documents
http://bfw.ac.at/iym/pdf/ziegner.pdf Accessed 17 May 2010 http://www.tirol.gv.at/leadmin/www.tirol.gv.at/themen/umwelt/wald/schutzwald/downloads/endber-kurzfass06-04-og.pdf Accessed 17 May 2010 http://www.us.is/id/4219 Accessed 17 May 2010 http://www.vedur.is/ofanod/log/ Accessed 17 May 2010 http://www.vegagerdin.is/storframkvaemdir/hedinsfjardargong/ 17 May 2010 http://www.vegagerdin.is/storframkvaemdir/oshlidargong/ Accessed 17 May 2010 http://www.vegagerdin.is/upplysingar-og-utgafa/aaetlanir/vegaaetlun/ Accessed 17 May 2010 http://www.vegagerdin.is/vefur2.nsf/Files/umferd_thjod_2008_skyrsla/$le/Umfer%C3%B0_%C3% A1_%C3%BEj%C3%B3%C3%B0vegum_2008.pdf Accessed 17 May 2010 http://www.vegagerdin.is/vefur2.nsf/Files/VegskraLysing/$le/Vegaskr%C3%A1_lei%C3%B0arl%C3%BDsing_ 01-03-2010.pdf Accessed 17 May 2010 http://www.vegagerdin.is/vegakerd/jardgong/jardgvegakerf/ Accessed 17 May 2010

Unpublished reports and diploma theses

, Fahrnberger W, Hallsteinsson H (2001) Siglufjo lafsfjo rurO rur. Veggo ng um Gumundsson A insfjo r. Yrlit yr jarfri Tro llaskaga og astur til jargangagerar. Jarfristofan ehf. He Unni fyrir Vegagerina tzGuer B (2007) Vergleichende Beurteilung des Lawinenrisikos auf den Verkehrswegen des hinteren O berlegungen zur Schulrelevanz. Diploma thesis, Institute of Geography, University of talsU Innsbruck Almenningum vi Siglufjo rsla vegna jarsigs a r Haiason H (1982) Jarfrisky

123

480

Nat Hazards (2011) 56:465480

Huttenlau M (2004) Risikoanalyse im Hinteren StubaitalTirol. Gefahrenprozess Lawine. Analyse des individuellen und kollektiven Todesfallrisikos auf der Zufahrtsstrae zur Talstation und Analyse des Schutzdezits im Bereich des Talstation der Stubaier Gletscherbahn. Diploma thesis, Institute of Geography, University of Innsbruck GVST, BAH Ra gjo ngsne tum r Ko f (1999) Greinarger um aurskriu u Jarfristofa A Fljo tursson HG et al. (2004) Kortlagning a s gi a Siglufjararvegi um Almenninga. Smundsson , Pe fangasky rsla A

123