Key points are not available for this paper at this time.
To find a link between microstructure and the mechanical properties of weak snow layers, we acquired a comprehensive dataset combining mechanical tests for compressive strength and stiffness on 278 weak layer samples with high-resolution 3D data from 70 co-located micro-computed tomography ( μ CT) scans. The samples encompass three main categories, often associated with dry-snow slab avalanche release: faceted crystals and depth hoar, decomposing particles and rounded grains, and buried surface hoar. The results for stiffness and compressive strength highlight different scaling behaviors with densities for different grain types, but also fundamentally different relations between stiffness and strength for persistent and non-persistent grain types. Using a neural network, we identified key clusters of microstructural features that are most predictive of mechanical behavior. These clusters were determined based on the intercorrelation of features and reflect different geometric and topological aspects of snow microstructure. Stiffness was primarily influenced by features describing the alignment and efficiency of load-bearing paths, such as anisotropy and tortuosity. In contrast, compressive strength was more sensitive to the local geometry of the ice-air interface, though it also depended on the load path characteristics. By distinguishing between these controlling factors, the model predicted the mechanical properties with an accuracy comparable to the experimental variance. While our results confirm that the mechanical properties primarily depend on density, they also demonstrate that snow microstructure shapes the specific relation between density and mechanical properties. This relationship differs markedly between persistent weak layers and those composed of decomposing particles and rounded grains.
Schöttner et al. (Sun,) studied this question.