Computer Tools from the SLF to Support Local Swiss. Avalanche and Road Officials. Michael Lehning, Walter Ammann, Perry Bartelt, Marc Christen, Martin ...
Computer Tools from the SLF to Support Local Swiss Avalanche and Road Officials Michael Lehning, Walter Ammann, Perry Bartelt, Marc Christen, Martin Gassner WSL, Swiss Federal Institute for Snow and Avalanche Research SLF, Flüelastrasse 11, CH – 7260 Davos Abstract This paper presents computer tools and models developed at the Swiss Federal Institute for Snow and Avalanche Research (SLF) and used by authorities in charge of avalanche safety in Switzerland. The tools are a) the statistical model of local avalanche danger NXDLawinen which searches a data base for similar situations in the past and presents the associated avalanche activity; b) the snow cover model SNOWPACK, simulating the detailed structure of the snow cover; c) the avalanche dynamics model AVAL-1D calculating runout distances and impact pressures associated with powder and dense-flow avalanches; d) the InfoBOX, an information system connecting specialists throughout Switzerland and collecting weather and snow data from different sources.
Figure 1: Large avalanche at the SLF test site Vallée de la Sionne.
Introduction About 65% (or 26'975 km2) of Switzerland’s area is alpine country. Approximately 1.7 million people (or 25% of the Swiss population) live in alpine regions and a number of important highways and railways cross the Alps. For example, over 19’000 vehicles per day cross the Gotthard, a very important transit route between Italy and Germany (Alpenreport 1998). Safety on the roads through the mountains is a particular challenge. For example, avalanche mitigation has always played an important role in the life of the people living in the Alps. Over the past 50 years about 1.7 billion Swiss francs have been invested in engineering work for avalanche protection such as snow supporting structures, deflectors or snow sheds. Together with avalanche hazard zoning this has led to a high degree of safety (compared to other hazards) in densely populated areas and on roads with high traffic volumes. Such permanent measures are not cost effective to protect recreational activities such as off-piste skiing and ski-mountaineering or less frequented roads, sparsely populated regions or ski areas. In addition, permanent measures can fail in extreme events, e.g. when large avalanches (Figure 1) hit roads or even freeways. Temporary measures such as evacuation of buildings during hazardous periods, temporary closure of roads or ski runs, artificial release of avalanches and careful route planning are both low cost and more flexible. However, they require an effective avalanche warning service supported by modern computer tools. Until the early 1990s avalanche forecasting was mainly based on intuition, experience and local knowledge of the forecaster. While these factors still play an important role today, progress in snow and avalanche research, especially in the development of computer models as well as rapid developments in sensor, communication and information technology during the last 5 to 10 years have opened up new ways in avalanche forecasting. Mathematical analysis of measurements, numerical simulations of weather and snowpack plus symbolic and statistical computations of the avalanche hazard are the key elements of modern avalanche forecasting, which can be described as „computer-aided avalanche forecasting”. In this paper, we present the new computer tools developed by the Swiss Federal Institute for Snow and Avalanche Research (SLF) and used by local avalanche and road officials. We also indicate, which tools could be useful for road authorities in other countries. NXD-Lawinen – A Forecast Tool for the Local Avalanche Danger NXD-Lawinen is a tool directly applicable to the problem of local avalanche forecasting. It assists the forecaster to take decisions such as “road or valley closure”, “village evacuation”, or “controlled avalanche release in certain paths”. The forecast tool is based on the nearest neighbour method (Buser, 1989; Gassner et al., 2000) and has been improved for more than two decades now. For each day, the system stores a set of snow and weather parameters (numerical values) together with the observed avalanche events. The information is stored in a data base initiated with known data from the past and grows as the system is operated. For the current day, when an avalanche danger forecast is required, the program searches the database for „similar cases“, using non-parametric statistics. With a suitable distance measure, the nearest neighbours can be calculated from the current weather and snow parameters. These parameters can be manually entered, or automatically retrieved from weather and snow stations or even weather forecast models. These new values are then stored and add another data set to the existing data, as soon as the result of the day, i.e. the information on observed avalanches, or non-occurrence of avalanches is additionally provided. The more data are available, the better the system works. NXD-Lawinen has a very user-friendly, graphical interface (Figure 2) and can be customized to the local peculiarities. A new feature is the possibility to enter observed
avalanches graphically. The quality of the forecasts increase with the time the system is in operation. In addition, the user has a tool helping him to increase his own experience and knowledge, because NXD-Lawinen serves as a companion, against which the judgements of the user are reflected. NXD-Lawinen is for sale from SLF.
Figure 2: NXD-Lawinen output showing the distribution of observed avalanches at different altitudes and aspects for all similar situations found for the current data set. The similar situations are marked on the table (left hand side). SNOWPACK – The Model of the Snow Cover SNOWPACK is a comprehensive model of the snow cover including modules for the underlying soil and drifting snow (Lehning et al., 1999). It simulates the evolution of the snow cover based on meteorological input data. International intercomparison studies show that SNOWPACK is successfully applied to alpine (Lehning et al., 2001), arctic, maritime and continental snow covers. SNOWPACK solves the mass- and energy balance equations using a Finite Element numerical scheme. It is written in the C programming language. SNOWPACK is delivered with a user-friendly Java interface. The interface is used to drive the model with key input parameters as well as to visualise the results. A variety of output and graphical options are available. SNOWPACK is available for windows and all UNIX systems. SNOWPACK parameterises snow microstructure and thus allows a detailed representation of the layered snow structure. The following individual processes are modelled: • Heat Transfer • Settling • Phase Change
• •
Water Transport Metamorphism
Figure 3: Comparison between modelled and measured grain type at Davos Versuchsfeld for the avalanche Winter 1999. The observed snow profile (right) is well simulated by SNOWPACK and important critical layers such as buried surface hoar are captured. The energy- and mass transfer at the surface is modelled by taking into account: • Precipitation as snow or rain • Shortwave radiation • Incoming and outgoing longwave radiation • Sensible heat flux • Latent heat flux including surface hoar formation • Windpumping • Snow drift SNOWPACK is developed primarily for the purpose of avalanche warning. It’s strength is the description of the snow cover layering and snow microstructure. Crucial weak layers and interfaces such as surface hoar, depth hoar or ice lenses are modelled (Figure 3). Because of its accurate mass- and energy balance, SNOWPACK is also increasingly used for climatological research. A new feature is the possibility to also model layers of soil or rock (to a variable depth). This is used for permafrost simulations. SNOWPACK has also a detailed description of the interaction with the atmospheric boundary layer. It treats complex processes such as wind pumping and snow drift (Lehning et al., 2000). Special attention is given to the treatment of shortwave radiation penetration the snow. Therefore, SNOWPACK is suitable for a detailed analysis of the energy- and mass fluxes between the atmospheric boundary layer and the cryosphere. This is applied to the reconstruction of the formation conditions of ice found in ice cores. The coupling of the SNOWPACK model with a snow drift model and an energy balance model is treated in a companion paper (Spreitzhofer et al.,
this issue). The possible application of the coupled model system to forecast snow distribution in complex terrain is also discussed in the companion paper. AVAL-1D – An Avalanche Dynamics Program for the Practice AVAL-1D is the new one-dimensional avalanche dynamics program that predicts runout distances, flow velocities and impact pressures (Figure 4) of both flowing and powder snow avalanches. Avalanche hazard maps are prepared by engineers and land-planners. These experts rely both on practical experience and calculation models. The one-dimensional computer-based models have now reached a level of development where they can be introduced to practitioners. Considerable snow avalanche know-how is still required to correctly use the program.
Figure 4: Example of simulation results showing flow velocities and avalanche dimensions for the Albristhorn avalanche in the canton Berne of Switzerland. Well calibrated depth-average continuum models are used to track the motion of the avalanches from initiation to runout (Figure 4). AVAL-1D consists of two computational modules: FL-1D dense flow avalanches (Bartelt et al., 1999; Sartoris and Bartelt, 2000) and SL-1D powder snow avalanches (Issler, 1998). Both are programmed in C. These modules solve the governing equations of mass, energy and momentum balance using an upwinded finite difference scheme. The graphical user interface was programmed in IDL (Interactive Data Language). The topography can be specified by hand or directly read from a digital elevation map. The avalanche and calculation parameters are entered using interactive dialogue windows. Either a dense flow or powder snow avalanche simulation can be performed. Results are
provided along the entire avalanche track, not only at specific points. Simulation output includes: • Runout distances and mass distribution in the runout zone • Animation of avalanche flow • XY-Plots at user-selected points along the avalanche track (time-history). • Profile-Plots at user-selected points (only for powder snow avalanches) • Possibility to overlay different simulation runs to visualize sensitivity studies The main area of application is hazard zoning, which includes the planning of roads and other infrastructure (Figure 5). But also permanent protection measures and possible worst case scenarios can be assessed with AVAL-1D.
Figure 5: Example of zoning in an Alpine valley with a main road running through the red zone. InfoBOX – Complete Information at Your Finger Tips While the internet is the main information channel for the general public, a service called InfoBOX is useful for professionals. This service links together national, regional and local avalanche and road specialists in Switzerland. The local road official needs a variety of information for his or her decisions. Such information includes weather forecasts, data from automatic weather and snow stations, the national and regional avalanche forecasts, and results from different computer model such as NXD-Lawinen or SNOWPACK, which is operationally run at the SLF. The InfoBox collects information from different sources such as the meteorological service, the SLF and the automatic stations and presents it in a condensed and clear form. About 150 specialists, i.e. people who are in charge of roads, ski areas or villages use the InfoBOX service. Via this service, snow and weather data from automatic stations or weather forecasts can be accessed 24h a day.
Conclusions The tools presented are used by local avalanche and road officials to assist in the decision process. This process is difficult because very often it needs to be a compromise between safety and economic gain. The tools and models discussed can not only be directly used to solve avalanche safety problems but might also be prototypes adaptable to other natural hazards such as rock fall, landslides and mud flows. In addition the numerical models of snow cover, energy balance and snow drift can be applied to forecast road conditions in snowy weather. More information on the tools and related subjects is available at our web site: www.slf.ch. Acknowledgments The development of the tools discussed has partly been funded by the Board of the Swiss Federal Institutes of Technologies, the Swiss National Science Foundation and the European Community. Many colleagues at SLF contributed to this work. We thank all of them and include our international partners. We especially thank Tom Russi for many ideas and some text. References: Alpenreport: Daten, Fakten, Lösungsansätze. CIPRA-International (Hrsg.), 1998. Bartelt, P., B. Salm and U. Gruber. 1999. Calculating dense-snow avalanche runout using a Voellmyfluid model with active/passive longitudinal straining. Journal of Glaciology, 45(150), 242254. Buser, O., 1989: Two years experience of operational avalanche forecasting using the nearest neighbours method. Annals of Glaciology, 13, 31-34. Gassner, M., K. Birkeland, Etter H.J. and T. Leonard. 2000. NXD2000: An Improved Avalanche Forecasting Program based on the Nearest Neighbor Method. In: Proceedings of the International Snow Science Workshop, Big Sky USA, 52-59. Issler, D. 1998. Modelling of snow entrainement and deposition in powder-snow avalanches. Annals of Glaciology, 26, 253-258. Lehning, M., Bartelt, P., Brown, R.L., Russi, T., Stöckli, U., Zimmerli, M., 1999: Snowpack Model Calculations for Avalanche Warning based upon a new Network of Weather and Snow Stations, Cold Reg. Sci. Technol., 30, 145-157. Lehning, M., Doorschot, J., Bartelt, P., 2000: A Snow Drift Index Based on SNOWPACK Model Calculations, Ann. Glac., 31, 382-386. Lehning, M., Fierz, C., Lundy, C., 2001 : An objective snow profile comparison method and its application to SNOWPACK. Cold Reg. Sci. Technol., in press. Sartoris, G. and P. Bartelt. 2000. Upwinded Finite Difference Schemes for Dense Snow Avalanche Modeling. International Journal for Numerical Methods in Fluids, 32, 799-821. Spreitzhofer, G., Lehning, M., Doorschot, J., Fierz, C., Raderschall, N., 2001. Integrated Model Approach to Forecast Snow Drift, Snow Cover and Surface Properties in Alpine Terrain and on Roads, Proceedings of the 11th SIRWEC2002 Sapporo, Japan, this issue.
