Regional Mountain Conference

Large-scale avalanche hazard maps are increasingly being used to identify at-risk areas and designate protective forests across mountainous regions. However, the accuracy of these automated approaches depends heavily on the quality of the input data, particularly the vegetation height models (VHMs) applied to characterise forest structure. The ability of current methods for generating VHMs — including photogrammetry, LiDAR and satellite imagery — to capture vegetation characteristics relevant for avalanche protection assessment, varies considerably. These characteristics include canopy density, gap width and tree height. While high-resolution photogrammetry and LiDAR data are available in well-surveyed regions such as Switzerland, other mountainous countries lack access to such detailed datasets. This makes satellite-derived VHMs a crucial alternative for global hazard assessment. These methodological discrepancies may lead to inconsistent definitions of forests with protective effects and propagate through numerical avalanche simulations. This work aims to evaluate and optimise VHM generation methods to improve the accuracy of large-scale avalanche hazard indication modelling and their resulting delineation of forests with protective effects, while also assessing the viability of satellite-based approaches in regions where survey infrastructure is limited.

VHMs derived from three acquisition methods — airborne photogrammetry, airborne LiDAR and satellite imagery — are compared, and their performance in capturing forest parameters critical for avalanche protection is assessed. Large-scale numerical avalanche simulations are conducted at two test sites in the canton of Grisons: a high-altitude coniferous forest and a mixed coniferous-broadleaf forest at lower elevations. Protective forest extents are computed for each VHM input and the resulting hazard maps are compared to quantify differences in avalanche runout zones and protective forest boundaries.

Preliminary analyses suggest that optimized, higher-resolution VHMs provide more accurate delineations of protective forests. By quantifying the differences in accuracy between the various methods, this work will provide guidance on the acceptable trade-offs when using satellite-derived VHMs in regions where data are scarce, thereby supporting a more effective management of natural hazards in mountainous regions.