Avalanche Guidelines

Transcription

1 improved Accessibility: Reliability and security of Alpine transport infrastructure related to mountainous hazards in a changing climate Avalanche Guidelines guide lines for snow avalanches hazard PP4 PP10 PARAmount is supported by means of the European Regional Development Fund (ERDF)

2 Table of contents Chapter 1: Introduction Hazard assessment Existing models... 5 Chapter 2: Methodologies on regional scale Snow avalanche hazard mapping with AvalforLIN Input data Description... 7 Chapter 3: Methodologies on detailed scale Aval 1D Description Preparation of input data Advantages/disadvantages Scientific ability criteria Usability Ramms 2D Description Preparation of input data Advantages/disadvantages Scientific ability criteria Chapter 4: Preliminary analysis Collection of historical, cartographic and climate data Meteo climatic analysis Geomorphologic analysis Historical analysis Detection of the release and run out areas Chapter 5: Future developments and improvements Limits of existing models and necessary improvements Experimental activities Laboratory experiments References PARAmount page 2 of

3 Table of figures Figure 1.1: snow avalanche at Val di Rabbi, Trentino, April Figure 1.2: Example of BUWAL matrix... 5 Tab.1: Classes of probability and intensity for the hazard assessment according to the Province of Trento... 5 Tab 2.1: Input data for AvalforLIN... 6 Figure 2.1: The main criteria for snow avalanche release areas determination using a raster DTM... 7 Figure 2.2: The snow avalanche Energy Line Angle concept... 8 Figure 2.3: Snow avalanches hazard mapping with AvalforLIN for the Italian case study Passo Rolle. The initial 1x1m Lidar DTM has been resampled to 20x20m. In red the release areas, in grey the runout zones and in blue the recorded past event data Figure 2 4: An example of validation of the results obtained with AvalforLIN using a real event which had occurred in another region than the one use for the for the calibration of the French model. In red the release are determined with AvalforLIN, in blue the travel path determined with AvalforLIN, in the circles the real zones (yellow = release area, green = stopping point)... 9 Figure 3.1: Scheme of the snow mass Figure 3.2: Scheme of the powder snow avalanches Figure 3 4: Example of hazard map fort he bed site Passo Rolle (Italy) Figure 4.1: historical data collected in the hazard assessment for Passo Rolle, Italy Figure 4 2: Slope map (Rolle Pass, Italy) Figure 4 3: Steepest slope direction map (Rolle Pass, Italy) Figure 5.1: Particles detected through the Voronoi methods Figure 5.2: Pampeago site, with the positions oft he targets and cameras PARAmount page 3 of

4 Chapter 1: Introduction Snow avalanches are complex natural phenomena that take place on the mountain areas, representing an important threat for the safety of people and common properties. They are caused by unstable snow, and usually happen after rapid changes in weather conditions (wind, temperature, snow, rainfall), which influence the stress and the cohesive features of the snow pack (see Figure 1.) The assessment of snow avalanche hazard, and more generally of hydro geological hazard, is the basis for a proper spatial planning, for a correct dimensioning of protection measures and for a good practice of prevention and risk management. Hazard maps should highlight areas that might be affected by the event and should give information about the relative level of danger. Figure 1.1: snow avalanche at Val di Rabbi, Trentino, April Hazard assessment Currently, the hazard definition is not univocal, and it is established by local administrations through implementation rules, both in Italy and in European countries. By now, the most widespread methodology in Europe for the hazard assessment is based on the BUWAL matrix (in Figure 1.2), which considers the dependence between the intensity of an event and its probability of occurrence. Local administrations can decide to modify the original BUWAL matrix by changing the threshold values needed to define intensity. Intensity is evaluated on the basis of the physical quantities that characterize a phenomenon. For snow avalanches the parameters are the snow height and the impact pressure, which depends on velocity and density of the snow itself. Therefore, intensity estimation is carried out by simulating the propagation of different events with different return periods. So far, a methodology for defining the return time of snow avalanches has not been established, since a systematic registration of past events for the statistical analysis is still lacking. Consequently, the return period of the precipitation, which generates the accumulation of snow that triggers the avalanche, is taken as the return period for the snow avalanche itself. PARAmount page 4 of

5 Figure 1.2: Example of BUWAL matrix In Tab.1. there is the classification for the probability of occurrence and intensity as reported by the regulations of the Autonomous Province of Trento, Linee guida metodologiche per la perimetrazione delle aree esposte al pericolo valanghe. Tab.1: Classes of probability and intensity for the hazard assessment according to the Province of Trento Classes of probability Low Tr = 300 years Medium Tr = 100 years High Tr = 30 years Classes of intensity Low P < 3 Pa Medium 3 <P < 15 Pa High P > 15 Pa 1.1 Existing models Since the early fifties several attempts have been made to develop simple models for snow avalanches as supporting tools in the hazard mapping activity. Up to now two different approaches can be distinguished among the existing models: Empirical approaches Dynamic approaches (or physical mathematical models). Empirical models process data from historical events on a statistical basis, without considering the physics of the problem. For this reason they are only able to estimate the run out distance of snow avalanches and are widely spread in the regional scale analysis. Dynamic models are based on a physical description of the events, and they can reproduce the evolution of snow avalanches through a set of mathematical equations, from the initiation to the run out, computing some important parameters, like the velocity, the impact pressure, the height of the flowing layer and its distribution. They are usually used in a local scale hazard assessment. PARAmount page 5 of

6 Chapter 2: Methodologies on regional scale Different numerical models aiming at reproducing the propagation of snow avalanches can be used to estimate the run out distance of a known mass of snow. Empirical models: are based on statistical elaborations of data coming from historical events. They derive the run out distance of an avalanche either from a regressive analysis (topographic/ statistical models) or from the nearest neighbours methodology (comparative models). Thanks to the concept of probability it is possible to quantify the uncertainty of the measurements by using only few input data. However they do not give information about dynamic quantities, and about the return period relative to known run out path for an avalanche. o Topographic/ statistical models: they were developed between the 70s and the 80s (Bovis e Mears, 1976; Lied e Bakkehoi, 1980) and are based on statistical regressions, which correlate the maximum run out distances of snow avalanches registered in different sites with their topographic characteristics. Another type of statistical models sort out the run out distance by inferring the distribution function of probability for a sample of data. In this case, data would come from different snow avalanches sites, or from the same site, where historical events are systematically registered, in order to have a correlation between the intensity of the events and their return period (McLung e Lied, 1987). o Comparative models: they work by averaging the extreme values of the run out distances of avalanches that have occurred in sites with similar topographic features to the site of interest. 2.1 Snow avalanche hazard mapping with AvalforLIN As RockforLIN, AvalforLIN is a 2D raster GIS model developed by Irstea. It allows natural risk experts to provide a snow avalanche risk assessment on a large geographical scale as the regional one. As for Rockfalls the input data necessary to use AvalforLIN are a DTM, the map of socio economic issues, the forest map and, if possible, the past event cadastre. It needs a field survey for its validation. Due to the fact that only topographic criteria are used and according to the DTM resolution and accuracy, some risk conditions can be under or over estimated. Only the confrontation with past events cadastre, and/or a campaign of field investigation can identify this over or under estimation. If this validation phase is not provided then the results obtain with AvalforLIN only produce a pre mapping which gives a first overview on the potential situation Input data Depending on the objectives, the input data needed for using AvalforLIN are the following ones written intab. Tab 2.1: Tab 2.1: Input data for AvalforLIN Objectives Release zones mapping Input data: raster maps (the resolutions can be different but the minimal resolution is the one of the DTM) Past event DTM cadastres or any Human geo localized infrastructures Forest map information PARAmount page 6 of

7 Run out zones mapping Risk mapping Protection forest mapping Description Snow avalanches release areas: AvalforLIN offers the possibility to perform a DTM analysis in order to identify the potential release areas on a large geographical sector, making the assumptions that they depend only on the topographic conditions and on the altitude of the case study. Slope thresholds can be applied to the slope surface raster. For snow avalanches the four criteria (Figure 2.1: The main criteria for snow avalanche release areas determination using a raster DTM) used for release area mapping are: the curvature, the slope, the altitude the surface. The thresholds of these criteria depend on regional and geo climatic conditions. Commonly in the European Alps, all cells in 25x25m raster DTM with a slope of degrees, a convex form, an altitude higher than 1000m and a minimal surface of 500m² are considered as potential release zones for avalanches. The resolution and the accuracy of the DTM affect the results. Z min = f(geo loc) Figure 2.1: The main criteria for snow avalanche release areas determination using a raster DTM Snow avalanches potential maximal run out areas AvalForLIN is based on the Energy Line Angle concept, initially formalized by Heim in 1932 and adapted with success to the snow avalanche context by Lied in As for rock falls, the most likely run out envelope is determined by the intersection between the ground and an imaginary line, which is drawn from the release point with a calibrated angle This angle ( ) is determined by means of an intermediary Energy Line Angle ( ), which goes from the highest point on the slope to the point where the deposition area starts. The former one is defined as the point for which the slope angle is equal to 10 (See Fehler! Verweisquelle konnte nicht gefunden werden.). The correlation between and, = a* + b, is determined using recorded and mapped past events. The calibration of this ( ) model needs to determine the travel path by using the principle of water PARAmount page 7 of

8 flow direction along the steepest slope. The model is not able to represent avalanche spread off in the run out zone. For this reason +, AvalforLIN uses a buffer zone around the maximum run out point. The width of this buffer zone depends on the local slope morphology. The maximal width of the buffer zone has been fixed at 220m. This value corresponds to the standard deviation (5.17 ) of calculated with the data set for the calibration of the regional models. The use of this buffer zone increases the confident interval of the models (See Figure 2.3). The robustness of the model calibrated using past events data depends on both the accuracy of the mapped past events envelopes and on the resolution of the DTM used for the data processing. Initial release point Maximal run out point Figure 2.2: The snow avalanche Energy Line Angle concept Validation of AvalforLIN The model AvalforLIN should be validated using for each case study some past events which have not nee used for the model calibration (see Figure 2 4). Snow avalanches protection forest mapping: Identification of the forest coverage, which potentially has a protection function, is obtained by combining the endangered items map with the forest cover map, and by selecting all forested areas located above an endangered item and on the associated release zones. This selection is provided automatically by AvalforLIN. This map of potential protection forest areas should be validated by a field survey. PARAmount page 8 of

9 Figure 2.3: Snow avalanches hazard mapping with AvalforLIN for the Italian case study Passo Rolle. The initial 1x1m Lidar DTM has been resampled to 20x20m. In red the release areas, in grey the run out zones and in blue the recorded past event data. Figure 2 4: An example of validation of the results obtained with AvalforLIN using a real event which had occurred in another region than the one use for the for the calibration of the French model. In red the release are determined with AvalforLIN, in blue the travel path determined with AvalforLIN, in the circles the real zones (yellow = release area, green = stopping point) PARAmount page 9 of

10 Chapter 3: Methodologies on detailed scale Physical models should be used in order to evaluate more precisely the intensity of snow avalanches. The regional scale analysis, in fact, identifies the areas that are more prone to the phenomenon, narrowing the field of investigation for the following study, but the intensity evaluation of the event can be done only by models that are physically based. Mathematical physical model can be classified on the basis of the type of snow avalanches that they are able to analyze. For dense snow avalanches, which are more relevant in the Alpine mountains in comparison with the powder ones, models can be subdivided in two classes: - Centre of mass models; - Continuous models. For powder snow avalanches there are: - Density current models, based on a single phase approach, which are similar to the ones used to study the turbidity of submarine debris flows; - Binary mixture models, based on a two phase approach, where the equation of mass and momentum balance are written both for the solid and the air components - Coupled models that combine features from the previous models, since they use different balance equation for the mass of snow and air, but only a momentum balance equation for the mixture. In the PARAmount project, hazard assessment was carried out through the Aval 1D and Ramms 2D, which are discussed in the following paragraphs. These two commercial codes used for the hazard assessment are based on the Voellmy Salm model (Salm et al., 1990) The avalanche motion is described only by the avalanche mass center, and that many approximations are introduced into the calculation method. Specifically: The flow behavior is represented only by the motion of the avalanche mass center. That is the mass distribution around the center of mass, or the internal deformation of the avalanche body, does not enter into the flow physics. The mass of the avalanche remains constant along the path. The three dimensional path is simplified into three segments identified along the main flow direction (x axis). The three segments represent the release zone, the flowing zone and the deposition zone. Release and flowing zone have constant slopes. Only in the deposition zone changes of slope angle can be considered. Each segment has a constant flow width. The equation of motion is solved analytically for each segment considering steady and uniform flow; moreover, the velocity U for each segment and run out distance are the main outputs of the calculations. The avalanche motion is controlled by the friction that is expressed by the sum of two contributions: a Coulombian term, dependent on the weight of the snow above, and a term proportional to the squared velocity like in the Chezy expression. 3.1 Aval 1D Aval 1D is based on the physical description of the dynamics of snow avalanches, and it uses a set of equations that give a reliable description of the phenomenon from the starting zones to the deposition area. In particular, It is able to predict some of the physical variables that are useful for hazard mapping, such as the run out velocities, the impact pressures, the flow level and the distribution of the deposit depth. The software forecasts dynamic variables both for dense snow avalanches (calculation module FL 1D) and powder snow avalanches (calculation module SL 1D ). Both of these modules solve the equations of mass and momentum conservation with finite difference methods. PARAmount page 10 of

11 3.1.1 Description FL 1D is a quasi one dimensional continuum model, in which an averaged velocity is defined along the depth (see Figure 3.1) The model is based on several important assumptions: Figure 3.1: Scheme of the snow mass. 1. Flowing snow is modelled as a fluid continuum with a mean constant flow density. 2. The flow width is known. 3. A clearly defined top flow surface exists. 4. The flow height is the average flow height across the section. 5. The vertical pressure distribution is hydrostatic. 6. Flow velocity and depth is unsteady and non uniform. 7. The avalanche mass is constant and no entrainment processes are modelled. The differential equations are solved numerically using first and second order upwinded finite difference schemes (see Sartoris and Bartelt, 2000). The model employs a Voellmy fluid flow law, in which the flow resistance of the Coulombian type is concentrated at the base of the avalanche. Figure 3.2: Scheme of the powder snow avalanches PARAmount page 11 of

12 SL 1D consists of a suspension layer and a so called saltation layer, according to Norem's description of powder snow avalanche formation and structure (Norem, 1995). The suspension layer has a depth of few meters and is modelled by depth averaged mass and momentum balances (see Figure 3.2). Mass and momentum exchange between the two layers is determined by particle settling, turbulent diffusion against the concentration gradient, and aerodynamic shear forces. The net erosion or deposition rate is a function of the kinetic energy of the impacting particles Preparation of input data As input data Aval 1 D needs the topography of the avalanche, its release conditions and the model parameters (coefficients of friction), which can be entered using dialogue windows, text files or even topographic maps. During the definition of the topography, some simple rules must be considered, as taking the real distance between two consecutive points within m, or limiting their difference in altitude to 20m. In case of sudden change in slopes, smaller distances can be specified in order to have a more accurate model of the real topography. As input data Aval 1D also requires the avalanche width, which can be derived from the observed avalanches and from preliminary studies, the release depth, which is function of the altitude, slope and climatic conditions. And the friction parameters. The latter ones are function of the geometry and the snow features, and are tabled in Aval 1D for different period of times. The values come from a thorough calibration carried out on several historical events. The model parameters can be changed and this allows the user to investigate the influence of topography and parameter variations on simulation results Advantages/disadvantages Strengths Weaknesses Opportunities Threats It requires a minimum of input data that can be easily obtained. The geometric data can be identified from GIS analysis and consist in coordinates of the points defining the path of snow avalanches. Snow data derive from climatic analysis, while frictional data are tabled values and they depend on the snow avalanche type and on topography. The output data (velocities, pressure, depth of snow deposit) are easy to be interpreted. The run out path of the snow avalanche is predetermined as input datum, like the stopping region and the flow depth. AVAL 1D is very sensitive to slope variations in the flowing region of snow avalanches. The snow avalanche mass is constant, since the mass variation processes are not modelled. Displaced volumes only depend on the height of the snow in the release area (evaluated thanks to the snow measurements available in the study area) and on the surface of the run out path. The application to regional or local scale affects only the accuracy of the preliminary analysis. Climate change can be taken into account by modifying the input parameters determined by historical and meteo climatic analysis. The hazard maps that have been produced in PARAmount is very similar to the CPLV (map of likely locations of avalanche based on the past events). The results from the AVAL 1D model might be less good If the CPLV is not available. The reliability of the results depends on the precision of the CPLV. PARAmount page 12 of

13 3.1.4 Scientific ability criteria Realistic results Results comparable to field observations Possibility of back analysis X Facility to insert barriers X X X Usability Applicability of parameter default Experience with the software* X Preparation of input data X Output (clarity and timing)** X *Experience could help user to make useful choices for more accurate outputs, like increasing the points that define the avalanche track in those areas where there are bends, gullies, change in slopes, ect. ** Aval 1D produces results in a reasonable amount of time X PARAmount page 13 of

14 Figure 3.3: Hazard map for the Baška grapa test bed, Slovenia, carried out with Aval 1D. The 20 km long section of the railroad through the test bed was divided into 10 meters long stretches. PARAmount page 14 of

15 3.2 Ramms 2D The software package RAMMS (Rapid Mass Movements), developed at the WSL Institute for Snow and Avalanche Research SLF, combines 2 dimensional process modules for snow avalanches, debris flow and rock fall together with a protection module SLF (e.g. forest) and a visualization module (GUI) in one tool. RAMMS is based on the Voellmy Salm avalanche dynamical model and a digital terrain model (DTM) Description RAMMS solve numerically the differential equations for the mass and momentum balance, together with the equation for the kinetic energy. All the variables are averaged along the depth and the integration is performed through the finite volume method, with an accuracy of the second order. The software is based on the mathematical model of Völlmy Salm. Although the VS model is able to forecast the velocity and the maximum height of the flowing layer at the front of the snow avalanche, it cannot predict the evolution of these parameters along the whole mass. For this reason RAMMS includes another model based on the production, transport and dissipation of the kinetic energy of the particle velocity fluctuations (RKE). The amount of the specific energy is correlated to the parameters of the Völlmy Salm model, which are considered variable along the snow avalanche profile Preparation of input data Like for AVAL 1D, RAMMS requires topographic input data (ascii format), project boundary coordinates and georeferenced maps or remote sensing imagery, which should be prepared in advance. Georeferenced datasets have to be in a Cartesian coordinate system, polar coordinate systems are not supported. it works with the digital terrain model, which helps users to have a realistic description of the topography. The definition of release areas and release heights have a very strong impact on the results of RAMMS simulations. In its most basic form, the Since RAMMS: Avalanche module employs the well calibrated Voellmy friction model, it contains two parameters: the Coulomb friction and the velocity squared dependent turbulent friction. These parameters can be selected as constant for the entire problem domain, or can vary spatially to account for variations in terrain characteristics, roughness or vegetation. Swiss guideline suggestions for friction parameters (based on extensive model calibration) are also available Advantages/disadvantages Strengths Weaknesses Opportunities Threats As input data it requires: the DTM along with the extension of the release areas; snow data, equal to those ones computed for AVAL 1 D; frictional data, which depend on physical features of the snow, on Mass variation processes are note modelled yet. Interactions between the bed and the snow avalanche body are not taken into account. It is less dependent on the topography of the study area. It can give information on the likely paths of snow avalanches by considering the kinetic energy owned by It could be very timeconsuming if it is applied to regional scale. The cost is higher than the AVAL 1D model. PARAmount page 15 of

16 initial volumes, type of snow covering and on topography. the mass. UnlikeAVAL 1D it does not need the run out path and its width, but only the extension of the domain Scientific ability criteria Realistic results X Results comparable to field observations X Consideration of protection forest X Possibility of back analysis X Facility to insert barriers X Usability Applicability of parameter default Experience with the software* X Preparation of input data X Output (clarity and timing)** X X *Experience could help user to make useful choices for more accurate outputs, like increasing the points that define the avalanche track in those areas where there are bends, gullies, change in slopes, ect. ** RAMMS 2D produces results in a considerable amount of time if DTM is very refined PARAmount page 16 of

17 Figure 3 4: Example of hazard map fort he bed site Passo Rolle (Italy) PARAmount page 17 of

18 Chapter 4: Preliminary analysis As pointed out several times in the previous paragraph preliminary analyses are very important in order to have very accurate and reliable results of the model. The preparation of a hazard map involves, in fact, some preliminary steps: 2 Review of the historical avalanche cadastre, with climatic and cartographic data.. 3 Determination of climatic conditions such as new snow depths, main wind directions and snow drift conditions. 4 Geomorphologic analysis. 5 Determination of avalanche type and return period through an historical study. 6 Detection of the release and run out areas 4.1 Collection of historical, cartographic and climate data They are essential for the site analysis and the determination of local climate conditions, like the fresh snow thickness and the direction of predominant winds, along with other factors responsible for large displacements of snow Figure 4.1: historical data collected in the hazard assessment for Passo Rolle, Italy 4.2 Meteo climatic analysis It is necessary to forecast the likely snow accumulations in the release area and the possible wind overloads. The study concerns data of: Temperature Rainfalls Snow precipitation Wind directions. 4.3 Geomorphologic analysis PARAmount page 18 of

19 It is carried out by analyzing existing geographic and cartographic information in order to identify the basins that might be affected by snow avalanches and to delineate more precisely the release areas, the flow and stopping regions. It includes: the analysis of the digital terrain model (DTM) to produce the slope map Figure 4 2: Slope map (Rolle Pass, Italy) The analysis of the DTM for the map of steepest slope directions Figure 4 3: Steepest slope direction map (Rolle Pass, Italy) The analysis of the ortophoto to assess the vegetation coverage. 4.4 Historical analysis It is based on the reconstruction of the main snow avalanche events that have affected the test bed site in the past. This type of analysis is useful to calibrate the numerical model used in the simulation of snow avalanche dynamics. If historical data and the map of likely locations of avalanche (CLPV) are not available, there is the PARAmount page 19 of

20 chance of making an analysis on a regional scale by using statistical models or other tools based on GIS operations, which can define the potential locations of snow avalanches, the release area and the flow path on the basis of slope and vegetation information. In this case the process is governed by the principle of the energy grade line. 4.5 Detection of the release and run out areas It is carried out by using GIS analysis through the support of the snow avalanches cadastre. The definition of the release area is conducted through GIS software, by elaborating the territorial data in thematic groups. The slope map The combination of the DTM and orthophotos leads to the definition of potential release areas, according to the criteria obtained by summarizing indications available in literature (David McClung e Peter Schaerer): 5 convergence towards a run out area that is documented (snow avalanche map); 6 slopes ranging between 30 and 50 ; 7 uniform vegetation of herbaceous or bushy type, with possible sparse wood; 8 ground with sufficiently uniform surface. The area identified in this way should be considered as the maximum that can be predicted for the avalanche site, while simulations can take into account sub areas of the same region. A careful analysis of the steepest slope directions allows the individuation of the run out profile, which is easy to define if there are marked beds. Chapter 5: Future developments and improvements 5.1 Limits of existing models and necessary improvements The most widespread commercial codes are not able to simulate entrainment processes and changes in the snow avalanche mass. Furthermore, they usually are employed over a fixed bed, excluding the contribution of snow from lower layers. Therefore, some improvements are needed in modeling snow avalanches, especially as concern: Deposition and erosion processes, which could define more realistically the depth of the snowcover in the deposition zone. Entrainment processes, which are responsible for the increase of the avalanche mass while it flows along its track. Recently some steps have been made to simulate these elements that influence avalanche speed and mass deposition. The RAMMS avalanche model is now being updated to include many of the results coming from theoretical and experimental research taken in Switzerland, and it is going to offer advanced model functions to specialized users, located primarily at the other research organizations. However, further studies on the rheology of snow avalanches are needed, in order to have a deeper understanding on the dynamics of snow grains forming the avalanches and model the interactions between the mass and the erodible bed. Most of all, field observations are required, since they could give important information about real events in their environment. Measurements carried out on artificial avalanches along canals are more frequent than survey carried out on full scale test site, especially because of the technical complexity in the monitoring activities and because of the high danger that is with these phenomena. In literature several studies are present on this topic, and an European project was focused on the realization of a monitoring site. PARAmount page 20 of

21 5.2 Experimental activities One of the PARAmount activity was the evaluation of future improvements of existing models, by beginning some experimental considerations, The Province of Trento has begun to arrange a test site in its territory in order to conduct an investigation on the dynamics of snow avalanches and the deposition process. The study has started with a preliminary analysis of the experimental conditions in laboratory, where the techniques that would be employed in test site have been examined. In the following paragraphs a short description of the activities are going to be presented Laboratory experiments The experimental apparatus was developed by the Laboratory of Hydraulics of the University of Trento in order to study both the dynamics of the avalanche and the impact pressure against a barrier. The system is made up of a short inclined channel having transparent side walls. It has a reservoir in the upper part, representing the release area, and an horizontal plane in the lower part, representing the deposit region, where a simple structure provided with a pressure sensor is installed in order to simulate a snow avalanche barrier. Two types of experiments were carried out: 1. The flow of the particles was filmed by the means of two synchronized high speed cameras, in order to reconstruct the 3 dimensional trajectories of the spheres across the channel. This activity was achieved through the use of a robust pattern matching algorithm based on the Voronoï diagram, which has been developed by the university of Trento. 2. The impact pressure of the avalanche against the barrier was filmed and registered through the pressure sensor, in order to have information about the velocity and concentration distribution near the obstacle, and to have an estimation of the entity of the force exerted by grains against protection barriers. 5.1: Particles detected through the Voronoi methods The first activity was useful to determine the number and the type of targets needed to perform videorecording in the test site and to understand the limits and the opportunities of the Voronoi techniques on real sites Field monitoring: Experience at Pampeago site This part of the study has been split in two parts: PARAmount page 21 of

22 11 Survey over the likely test sites in Trentino, in order to find the most feasible in terms of usability of the imaging techniques that would be used, of possibility of triggering artificial avalanches and in terms of safety conditions during the measurement campaign. 12 Installation of the instruments defined in the preliminary investigation. After a thorough investigation, the site of Pampeago was chosen as the best candidate. The positions of targets and of a firm supporting for the acquisition with laser scanner have been defined by analyzing the maps and the ortophotos of the area. The instruments was planned and produce by the Laboratory of Hydralucis of the University of Trento, in collaboration with the Servizio Geologico of protezione Civile of PAT For the bearing of cameras and laser scanner, three stable supporting have been installed on the opposite side of the site. 16 targets on 8 sticks provided with stay rodes have been put along the channel, out of the run out track but inside the area filmed by the cameras. 5.2: Pampeago site, with the positions oft he targets and cameras PARAmount page 22 of

23 References [1] BOVIS E MEARS, 1976, Statistical Prediction of Snow Avalanche Runout from Terrain Variables in Colorado, Artic and Alpin Research 8, [2] LIED E BAKKEHOI, 1980, Empirical Calculations of Snow Avalanche Runout Distance Based on Topographic Parameters, Journal of Glaciology, 24(94), [3] MCCLUNG E LIED, 1987, Statistical and Geometrical Definition of Snow Avalanche Runout, Cold Regions Science and Technology, 4, [4] PERLA, CHENG, MCCLUNG, 1980, A Two Parameter Model of Snow Avalanche Motion, Giournal of Glaciology 13, [5] NOREM, IRGENS E SCHIELDROP, 1989, Simulation of snow avalanche flow in the runout zones., Annals of Glaciology 13, [6] NAAIM, 1992, Modelisation of Dense Avalanches, Proc. of the European Summer University on Snow and Avalanches, September 1992, Chamonix, France, Cemagref Publications, [7] BARBOLINI, M., GRUBER, U., KEYLOCH, C., NAAIM, M. E SAVI, F., 2000, Application and evaluation of statistical and hydraulic continuum dense snow avalanche models to five real European sirtes., Cold Regions Science and Technology, 31(2), [8] M. CHRISTEN, P. BARTELT AND U. GRUBER, 2002, AVAL 1D: an avalanche dynamics program for the practice, SLF Davos [9] G. SARTORIS AND P. BARTELT, 2000, Upwinded finite difference schemes for dense snow avalanche model, Swiss Federal Institute for Snow and Avalanche Research [10] P. BARTELT, B. SALM, U. GRUBER, 1999, Calculating dense snow avalanches runout using a Voellmy fluid model with active/passive longitudinal straining, SLF, Davos [11] M. CHRISTEN, Y. BUEHLER, P. BARTELT AND L. SCHUMACHER, March 2010, Ramms1.3.0 User Manual, WSL Institute for Snow and Avalanche Research SLF [12] M. CHRISTEN, P. BARTELT, J. KOWALSKI, L. STOFFEL, 2008, Calculation of dense snow avalanches in threedimensional terrain with the numerical simulation programm Ramms, WSL Institute for Snow and Avalanche Research, SLF, Davos [13] F. BERGER, 2012, Rockfalls and snow avalanche hazard mapping using the Energy Line Angle concept, IRSTEA PARAmount page 23 of