Dynamic crack propagation in weak snowpack layers: insights from high-resolution, high-speed photography

Abstract

Abstract. Dynamic crack propagation in snow is of key importance for avalanche release. Nevertheless, it has received very little experimental attention. With the introduction of the propagation saw test (PST) in the mid-2000s, a number of studies have used particle tracking analysis of high-speed video recordings of PST experiments to study crack propagation processes in snow. However, due to methodological limitations, these studies have provided limited insight into dynamical processes such as the evolution of crack speed within a PST or the touchdown distance, i.e. the length from the crack tip to the trailing point where the slab comes to rest on the crushed weak layer. To study such dynamical effects, we recorded PST experiments using a portable high-speed camera with a horizontal resolution of 1280 pixels at rates of up to 20 000 frames s−1. We then used digital image correlation (DIC) to derive high-resolution displacement and strain fields in the slab, weak layer and substrate. The high frame rates enabled us to calculate time derivatives to obtain velocity and acceleration fields. We demonstrate the versatility and accuracy of the DIC method by showing measurements from three PST experiments, resulting in slab fracture, crack arrest and full propagation. We also present a methodology to determine relevant characteristics of crack propagation, namely the crack speed (20–30 m s−1), its temporal evolution along the column and touchdown distance (2.7 m) within a PST, and the specific fracture energy of the weak layer (0.3–1.7 J m−2). To estimate the effective elastic modulus of the slab and weak layer as well as the weak layer specific fracture energy, we used a recently proposed mechanical model. A comparison to already-established methods showed good agreement. Furthermore, our methodology provides insight into the three different propagation results found with the PST and reveals intricate dynamics that are otherwise not accessible.

Crack propagation speeds in weak snowpack layers

“…Since similar experimental crack speeds were observed throughout the three crack propagation events, it can be assumed that crack propagation processes observed in the PST were similar to those in the whumpf and the cross-slope propagation in the avalanche. The increase of crack speed in the first 2 m of the PST beam highlights the need for long PST experiments, at least longer than the so-called touchdown distance, which is typically 2–3 m (Bair and others, 2014; Bergfeld and others, 2021b). If this is fulfilled, our results suggest that crack propagation in a PST is representative for larger 2-D crack propagation occurring in whumpfs and for the cross-slope direction in our avalanche movie.…”

“…of the noise in z -displacement before crack propagation) to the z -displacement to locate the crack tip at each time step (video frame) to derive crack speed, as the slope of linear fits in beam sections, along the PST (beam section width: 50 cm, step size: 5.3 cm). A detailed description of the applied methodology, including uncertainty estimates, is given in Bergfeld and others (2021b).…”

“…Although values obtained with the DEM model underestimated crack speed compared to the experiment, the MPM model overestimated crack speed. Away from the PST beam ends (2 m < x <4 m), where edge effects are small (Bergfeld and others, 2021b), mean speeds were 44 ± 6, 31.6 ± 2.1 and 63 ± 3 m s −1 for the experiment, DEM and MPM, respectively.…”

“…A high-speed camera (Phantom, VEO710) was used to record the speckled wall with an image resolution of 1280 pixel by 352 pixel at a rate of 7000 frames per s (lens aperture: f 2.8, focal length: 24 mm). Camera distortion correction and DIC analysis were performed as described in Bergfeld and others (2021b).…”

“…Reported crack speeds range from 10 to 50 m s −1 (van Herwijnen, 2005; van Herwijnen and Birkeland, 2014; van Herwijnen and others, 2016b). Recently, high-resolution crack speed estimates were obtained with a digital image correlation (DIC) method, highlighting variations in crack speed along a PST and pronounced edge effects (Bergfeld and others, 2018, 2021b).

Fig.

…”