ID25: Gravitational mass flow simulations for avalanches

Abstract ID 388 | Date: 2022-09-12 10:09 – 10:18 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Stoffel, Lukas (1); Christen, Marc (1,2); Bühler, Yves (1,2); Stefan, Margreth (1); Bartelt, Perry (1,2)

In Switzerland extreme value statistics is used to define snow fracture heights for practical avalanche dynamics calculations. The method relates measured three-day snow depth increase to return period at representative snow observation stations. The fracture heights are then modified to include terrain effects (primarily steepness), elevation and snow drift. The procedure is an integral part of avalanche hazard mitigation in Switzerland, especially since it allows engineers to study 30Y, 100Y and 300Y avalanche scenarios using a consistent and observation-based method.

The important role of snow entrainment in avalanche dynamics calculations is well known. Not only can entrainment change the mass balance of a specific event, it also influences the overall avalanche flow regime (e.g. powder or wet). Advanced avalanche dynamics models consider entrainment, including the temperature and moisture content of the entrained snow. Flow rheology is made temperature and moisture dependent, including fluidization and lubrication processes which lead to different avalanche flow regimes. The problem remains, however, of how to consistently define the snowcover disposition, temperature and moisture content in complex terrain, for different climatic regions as input for simulations.

In this paper we determine the snow entrainment heights using extreme value statistics. Similar to the case of fracture heights, the entrainment heights are related to measured three-day snowfall amounts. The procedure uses the same representative snow observation stations and therefore can be integrated easily into existing calculation methods. Snow heights are altered to include elevation and temperature gradients. However, the effects of snow drift are only included in the determination of fracture heights. We apply the method to several recent avalanches where the mass balance has been captured with aerial drones. Because the events are recent, we have reliable temperature estimates. Although the method appears to be applicable for the 30Y avalanches we investigate, we express our concerns regarding extreme events where snowcover distribution and temperature might vary strongly from the proposed procedure. Finally, we discuss how the method could be improved and expanded to investigate climate change scenarios on future avalanche activity.

Abstract ID 556 | Date: 2022-09-12 10:18 – 10:27 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Wirbel, Anna; Tonnel, Matthias; Neuhauser, Michael Johannes; Oesterle, Felix; Fischer, Jan-Thomas

Simulation tools for snow avalanches are often applied in a deterministic manner, i.e. yielding a single simulation result. However, numerical modelling results have inherent uncertainties originating from different sources. For one, due to limitations in process understanding as well as constraints on available input data, the chosen set of equations itself might not fully represent the physical processes at work. When solving the mathematical model with numerical methods, simulation results provide only an approximation to the true solution, based on the type of method, associated parameters and its implementation. Furthermore, data on the initial state or boundary conditions can suffer from uncertainties which propagate through the model.

Performing probabilistic simulations is an approach to quantify uncertainties in modelling results and also to asses the range of uncertainties introduced by different sources.

Here we present an uncertainty assessment of avalanche simulations using the open-source framework AvaFrame. We analyze uncertainty introduced by two different sources: a) a parameter related to the numerical implementation and b) the uncertainty in input data propagating through the numerical model.

For this purpose, we employ the dense flow snow avalanche simulations module (com1DFA) and vary the initial particle distribution (case a) and the release snow thickness (case b), which mainly define the avalanche release. Uncertainties in the resulting flow variables are quantified using the statistical module (ana4Stats) and visualized with probability maps. Converting the simulation results into an avalanche path following coordinate system (ana3AIMEC module), allows us to compare and to determine the likely range in scalar indicators like peak pressure-based runout length or maximum peak flow thickness and velocity. Combining these analysis tools, we can also compare the magnitudes of uncertainties introduced by different sources.

On the basis of these two examples, we showcase how AvaFrame can be used as a tool i) to study the effect of variations in numerical and input parameters and the resulting uncertainty bounds as well as ii) to include information on uncertainty within simulation result visualisations.

Abstract ID 165 | Date: 2022-09-12 10:27 – 10:36 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Bühler, Yves (1,2); Stoffel, Andreas (1,2); Christen, Marc (1,2); Margreth, Stefan (1); Stoffel, Lukas (1); Marty, Christoph (1); Bebi, Peter (1,2); Kühne, Roderick (3); Bartelt, Perry (1,2)

Snow avalanches threaten people and infrastructure in alpine regions around the globe. In order to be able to assess the avalanche risk for these infrastructures, hazard mapping procedures have been developed and successfully applied. They combine avalanche history, terrain analysis, field investigation and snow climatic information with numerical simulations. Avalanche hazard maps show the hazard on a parcel-by-parcel scale and are used for land use planning purposes. In Switzerland, four different zones are defined based on the frequency and pressure of an avalanche. In the area of highest hazard (red), it is forbidden to build houses; in the moderate hazard zone (blue), houses are allowed to be built, but have to be reinforced to withstand impact pressures of up to 30 kPa. But this costly procedure is only elaborated where large values are at risk. In all other alpine regions (90 % in the case of the canton of Grisons), avalanche hazard is not systematically assessed.

To close this gap, we developed an automated approach to generate spatial continuous hazard indication mapping based on digital elevation models. This enables the calculation of different scenarios with return periods ranging from frequent to very rare as well as with and without taking the protective effects of the forest into account. This approach combines the automated delineation of potential release areas, the estimation of release depths and the numerical simulation of the avalanche dynamics, applying the well-established RAMMS model, which is applied for hazard mapping in Switzerland and further countries. This procedure can be applied worldwide, where high spatial resolution digital elevation models, detailed information on the forest and data on the snow climate are available. It enables reproducible hazard indication mapping also in regions where no avalanche hazard maps yet exist, which is invaluable for climate change studies. In this contribution we give an outlook on already performed and planned applications in different countries on various continents.

Abstract ID 392 | Date: 2022-09-12 10:36 – 10:45 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Panayotov, Momchil (1); Tsvetanov, Nickolay (1); Bebi, Peter (2)

Pirin Mountains in Bulgaria are the refuge of some of the best-protected endemic Pinus peuce and Pinus heldreichii forests in the world. These species are famous for their longevity with many trees reaching more than 500 years. Due to the steep and long mountain slopes the forests are affected by avalanches and some of the trees keep record of past avalanche activity in their tree rings.

In our study we use combination of tree-ring analysis, satellite images and historical aerophotos to evaluate the effects of avalanches on forests in the Bunderitsa valley. We have collected tree-ring cores from affected trees on the borders between the forest and some of the large avalanche couloirs on the northwestern slope of Todorka peak and the eastern slope of Palashitsa and Vihren peaks thus collecting data on past avalanches with bigger sizes.

Our findings show that avalanches are the main shaping factor for the structure of forests in the valley followed by fires. Past avalanche activity has opened long-lasting avalanche tracks in the forests. About 56% of the potential forests (i.e. territories below treeline, outside of avalanche couloirs, streams, rock formations and screes) are strongly affected by avalanches. Of them 39% are in avalanche runout zones, 12% are in avalanche tracks in the forests and 48% are forests, which are periodically strongly affected by bigger avalanches. Comparisons with older aerophotos (1970s) showed that back then there were larger openings in the forests due to high frequency of avalanches in the very snowy 1950s and 1960s. Although recent avalanche activity has decreased, there are still years with larger events, which affect strongly forests. In addition, tourist development in vicinity has increased risk for human health and life due to avalanche accidents, including in forests.

By use of tree-ring analysis we reconstructed past avalanches that affected certain areas of the study region. The big couloirs are affected by smaller avalanches almost annually, while bigger avalanches have hit the neighboring forests almost every decade. By use of simulations with the RAMMS software and comparisons with recent records we evaluate the potential size and parameters of old avalanches. Our findings demonstrate that avalanches in the valley are of high importance and require more attention by authorities both as risk factor for human health and life and as natural disturbance shaping the forest structure and dynamics.

Abstract ID 658 | Date: 2022-09-12 10:45 – 10:54 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Schwarz, Jakob (1); Reiweger, Ingrid (1); Tollinger, Christian (2); Granig, Matthias (2)

For more than two decades the avalanche simulation program SAMOS/SamosAT (Snow Avalanche Modelling and Simulation) is in use by the Austrian Torrent and Avalanche Control. The program allows the simulation of dense flow avalanches as well as powder snow avalanches. SAMOS is well calibrated for big avalanches. However, simulations of avalanches smaller than 60,000 m³ often show too large run-out distances compared to documented avalanche events. One of the main parameters significantly influencing the run-out distance is the friction parameter µ. In SAMOS and the updated versions of it (Samos2007, SamosAT and AvaFrame) the friction parameter was created by various physical approaches to form the SAMOS friction model. The problem of overly wide-spread run-out distances coupled with the application of a single friction value for all avalanches suggests the need for more precisely defined parameters. To tackle this, we collected well-documented small (< 25000 m³) and medium (25000 – 60000 m³) avalanche events with at least one known point of the maximum run-out distance. Each of these avalanche events were simulated with a range of µ values to subsequently find the simulation that is most similar to the documented avalanche in terms of the run-out distance. To find the most suitable simulation, two different evaluation methods are used: (I) an assessment by experts as a reference and (II) a user-independent method for estimating the simulation results. The maximum runout in the evaluation I and II is always placed in relation to the 1 kPa outline of the simulation. Based on the characteristics of the avalanche simulation (lengths and widths per variation of µ), a set of predefined polygons is constructed. In the case that a simulation in the run-out reaches the area of the documented event, the simulation overlaps the predefined polygons. Already changes of one grid cell (5 x 5 m) shifts the absolute quantity of cells in the respective polygon. Thus, the method using the predefined polygons (II) proves to be very sensitive to the most minor changes.

Finally, the two evaluation methods are compared and a friction value µ, depending on the size of the avalanche and on the altitude of the run-out is defined. This allows a generally more accurate simulation of dense flow avalanches in SAMOS depending on their size.

.

Abstract ID 560 | Date: 2022-09-12 10:54 – 11:03 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Tonnel, Matthias (1); Von-Busse, Marie (2); Wirbel, Anna (1); Oesterle, Felix (1); Fellin, Wolfgang (2); Fischer, Jan-Thomas (1)

Validating the numerical implementation of process based gravitational mass flow models is a challenging but crucial step. Test cases with an analytical solution are rare. One example is the dam break case, a test case with depth integrated equations and Riemann initial conditions, which can be used to validate the first few seconds of a simulation. Another example is the similarity solution which allows to check the pressure gradient or the friction force implementation.

To extend the repertoire of precise reference solutions, we propose the use of a geometrical solution that is related to the total energy of the system. Soley considering coulomb friction this solution is motivated by the first principle of energy conservation along a simplified topography. Here friction force only depends on the slope angle. The analytical run-out is the intersection of the path profile with the so-called alpha line defined by the friction angle. It is also possible to extract from this alpha line information about the flow mass averaged velocity at any time or position along the path profile.

We present an implementation of this energy evaluation and verification approach in the simulation toolbox AvaFrame, including the following steps: the avalanche path profile is extracted from the mass average particles position from the Dense Flow Avalanche (DFA) simulation module (com1DFA) using Coulomb friction. On this path profile, the alpha line solution (which corresponds to the energy line) is computed and used as reference to validate the DFA simulation. Both velocity and run-out can be verified with this method.

We finally explore and explain the limitations of this approach. For example, the geometrical energy solution appears to only coincide with the DFA solution when the flow direction is in line with the steepest descent.

Abstract ID 527 | Date: 2022-09-12 11:03 – 11:12 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Li, Xingyue (1,2); Sovilla, Betty (3); Ligneau, Camille (3); Jiang, Chenfanfu (4); Gray, Nico (5); Gaume, Johan (2)

Erosion significantly affects the dynamics of gravity-driven mass flows and may lead to the formation of roll waves and/or erosion-deposition waves. Such wave phenomena can lead to severe flow variations and large flow depth, which have been one of the key concerns in many artificial and natural mass flows. Compared to roll waves which have moving material between the wave crests, erosion-deposition waves have completely stationary regions between individual crests. Although frequently observed from experiments and field events, roll waves, erosion-deposition waves and their transitions are challenging to be modelled under real-scale conditions involving complicated terrain and flow dynamics. Here, we study various erosion and entrainment behaviors as well as wave phenomena in snow avalanches using the material point method (MPM), finite strain elastoplasticity and critical state soil mechanics. With varied snow properties, distinct erosion patterns are obtained and analyzed with the mass change rate. When there is significant eroded and entrained mass, properties of released snow and erodible bed snow have clear correlations with the eroded mass, but not with the entrained mass, disclosing the difference in erosion and entrainment. Both enhanced and inhibited avalanche mobilities due to erosion and entrainment are captured under different conditions of snow properties and lengths of release and erodible zones. Furthermore, a snow avalanche at Vallée de la Sionne (VdlS) in Switzerland is modelled with consideration of bed erosion. The properties of the simulated snow are firstly calibrated with the deposition depth of the VdlS avalanche measured with a laser scanner. The dynamic behaviour of the avalanche is then analysed in terms of the different waves, the flow depth evolution at a fixed location, as well as the temporal and spatial evolution of the flow velocity. It is observed that both roll waves and erosion-deposition waves are naturally captured from the simulated avalanche. The flow depth evolution from the simulation shows satisfactory agreement with the field data measured with FMCW radar. The model is also able to recover the spatial evolution of the wave activity from release to deposit, which was measured in the field experiment using a radar. With both the numerical and field investigations, this study offers new perspectives on wave behaviour and provides a validated numerical approach for exploring waves in granular flows like snow avalanches. It may also stimulate the development of advanced erosion and entrainment models for large-scale avalanches.

Abstract ID 531 | Date: 2022-09-12 11:12 – 11:21 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Köhler, Anselm (1); Wirbel, Anna (1); Winkler, Michael (1,2); Fellin, Wolfgang (2); Oesterle, Felix (1); Fischer, Jan-Thomas (1)

Dynamic mass flow simulations provide multi-dimensional results of the spatio-temporal evolution of flow variables such as flow velocity, flow depth and derivatives thereof like impact pressure and associated run out. The analysis of simulation results is often reduced to final or peak states, e.g. comparing deposit outlines to flow depths of the last timestep, however, the dynamic aspects and the flow evolution over time are lost. This is reasonable from a historical perspective considering post event observation as damage or avalanche deposits are the main data source for model evaluation.

Recent studies using radar measurements acquire truly dynamic data of snow avalanche flows, offering the possibility to evaluate numerical simulation tools with respect to the temporal evolution of flow parameters. Radar data are typically displayed in a range-time diagram along the antenna line of sight and the distance to the approaching avalanche is shown for each time step. Converting the results of avalanche flow computation into simulated radar images using the radar coordinate system yields a synthetic range-time diagram that facilitates a direct comparison of radar measurements and modelling results.

This can be done not only from a radar’s point of view, but also along the thalweg – a 1D representation of the avalanche path. The coordinate projection onto a regular non-uniformly spaced grid with curvelinear coordinates along the thalweg reduces the spatial complexity of the simulated flow variables and enables a comparison between different simulation scenarios and different models. Such a reduction is known from systematic approaches to objectively evaluate and compare simulation tools, which are implemented in the open-source AvaFrame simulation toolbox.

We present a new thalweg-time (TT) diagram to assess and visualize the temporal evolution of the simulated flow variables and at the same time provide information on overall runout distance and the frontal approach velocity. Both representations of the simulation results, the map-like snapshot at run-out and the radar-like range-time TT-diagram greatly complement themselves. The TT-diagram adds the temporal evolution to the established spatial extent in the visualization and analysis of a simulated mass-flow event and provides a reference system for systematic model evaluation and comparison.

Abstract ID 200 | Date: 2022-09-12 11:21 – 11:30 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Jarosch, Alexander H. (1); Jóhannesson, Tómas (2)

Snow- and landslides pose a significant threat to settlements in avalanche-prone areas, where protective measures are widely used to improve safety. Simulations of the flow of avalanches against obstructions is essential for the design of catching and deflecting dams as well as other protective measures in run-out zones. Hazard zoning below protective measures and, in general, for avalanche paths with complex terrain geometry also requires advanced simulations of snow-avalanche dynamics . We present an efficient implementation of an incompressible granular-flow rheology for computational fluid dynamics (i.e. OpenFOAM), based on recent advances in the theory of μ(I) granular rheologies. Model parameters have been calibrated with observations from eight large Icelandic avalanches. Our simulations reproduce observed shapes of avalanche deposits in the run-out zones and available radar measurements of avalanche velocities. Observed and estimated values for avalanche volume and release depth in starting areas serve as initial conditions for our simulations. Several studied examples involve flow paths with complicated geometries, including deep gullies and ridges that split the simulated avalanche in the run-out zone, which indirectly provides constraints on the simulated flow dynamics.

Our approach represents an important improvement with respect to existing, depth-averaged models for snow-avalanche flow in complicated terrain geometries as it is able to simulate the full three-dimensional flow at impact with obstacles such as catching and deflecting dams and braking mounds. Our simulations are able to recreate the formation and time-dependent development of hydraulic jumps. Thus, splashing is formed at impact with obstacles as well as granular wedges behind the upstream face of dams or mounds alongside many other flow features expected in such complex granular flows. We will present several simulations of avalanche flows against protective dams in Iceland.

Abstract ID 499 | Date: 2022-09-12 11:30 – 11:39 | Type: Oral Presentation | Place: SOWI – Lecture hall HS1 |

Cicoira, Alessandro (1,2); Blatny, Lars (2); Li, Xingyue (2); Trottet, Bertil (2); Gaume, Johan (2,3)

Alpine mass movements can generate process cascades involving different materials including rock, ice, snow, and water. Numerical modelling is an essential tool for the quantification of natural hazards, but state-of-the-art operational models reach their limits
when facing unprecedented or complex events. Here, we advance our predictive capabilities for process cascades on the basis of a three-dimensional numerical model, coupling fundamental conservation laws to finite strain elastoplasticity. Through its hybrid
Eulerian-Lagrangian character, our approach naturally reproduces fractures and collisions, erosion/deposition phenomena, and multi-phase interactions, which finally grant very accurate simulations of complex dynamics. Four benchmark simulations demonstrate the physical detail of the model and its applicability to real-world full-scale events, including various materials and ranging through four orders of magnitude in volume. In the future, our model can support risk-management strategies through predictions of the impact of potentially catastrophic cascading mass movements at vulnerable sites.