Snow avalanches are a natural hazard exposing humans, infrastructure and landscapes to considerable risks. One crucial element of avalanche research is the size classification of avalanches. Providing an approach to evaluate the possible extent of avalanche events, in a quantitative manner, adds an improvement in avalanche simulations and forecasting hazardous areas. Therefore, this thesis develops and applies a statistical analysis of geodata in order to classify thalwegs as representation of all potential avalanche events within one path. It provides an analysis for spatial thalweg characteristics derived from geometric data based on terrain features as well as for intensity thalweg characteristics by incorporating the energy conservation principles. The dataset used as basis for the quantitative classification scale is derived from the Avalanche Modelling Atlas, a database hosted by the Austrian Research Center for Forests (BFW) and includes thalweg data from multiple sources. To analyse the data in a statistical and quantitative manner the avalanche framework AvaFrame was extended. Thereby, known formulas and calculations were adapted and have been implemented into the corresponding tools of the open source framework. The derived thalweg characteristics are divided into two categories, the spatial characteristics, including travel length Δs, altitude difference Δz and travel angles γ(s), as well as the intensity characteristics, including maximum velocity vmax, destructiveness Pmax and travel time tD. Furthermore, a maximum potential avalanche size is introduced as an attribute for each thalweg. It assigns each thalweg a class, according to the maximum potential avalanche event that may occur on a specific path. To evaluate the newly retrieved classification scale for thalwegs, the results were put in context with international classification schemes, namely the destructive size of the Canadian Avalanche Association and the avalanche size classification of the European Avalanche Warning Services. For the travel lengths, the three classification schemes show good alignment, with travel lengths between 50m and 200m for size two to 2000m and more for size five. However, the Canadian Avalanche Association mostly provides travel lengths on the higher end of the classes. For the destructiveness the here determined values suggest higher impact pressures for smaller avalanches (43 kPa for size two and 113 kPa for size three) as so far assumed and less destructive force for larger avalanches (317 kPa for size four and 616 kPa for size five). Furthermore, this approach was also tested on simulated thalwegs and events resulting in an overall well agreement between the different schemes with better alignment for large and extremely large avalanches and potential for improvement for smaller avalanche events. Some limitations such as the subjectivity within the digitising process and the simplicity of the model for the intensity characteristics highlight the importance of cautious interpretation and potential areas for methodological improvement. In conclusion, this thesis establishes the foundation for further advancement in the statistical analysis of geodata, offering a more objective, reproducible, and applicable methodology for classifying thalweg characteristics quantitatively.
