High-speed imaging of snow saltation: wind tunnel experiments using natural snow

Drifting snow is a multi-scale process. It is composed of particles rolling and sliding along the surface, particles in saltation following short ballistic trajectories in the first 10 cm above the surface and particles in suspension at higher regions of the atmosphere. Drifting snow is currently represented in some regional and mesoscale atmospheric models by taking into account its effect on snow height, snow sublimation and snow densification. Snow saltation is a sub-grid process in these models and is therefore parameterized. However, the current parameterizations are based on limited field and wind tunnel measurements and do not take into account the effect of the bed characteristics, as grain size, inter-particle cohesion and snow density, on the saltation dynamics. In order to improve the current saltation models, we conducted wind tunnel experiments using natural snow at the WSL Institute for Snow and Avalanche Research SLF to measure the kinematics and shape of particles in saltation. The wind tunnel is located at 1670 m above sea level, has a cross section area of 1x1 m2 and a total length of 14 m. Naturally deposited snow is collected in trays after each snowfall and transported to the tunnel without disturbing the snowpack. We used a high speed camera, aquiring images at 5 kHz with backlighting provided by an LED to capture images of saltating snowflakes. We measured wind speed with an array of pitot tubes positioned 2-10 cm above the snowbed. We additionally measured the density and hardness of the snow cover before the experiments using a box density cutter and a Snow Micro Pen (SMP), respectively. We process the images with a 2D Particle Tracking Velocimetry (PTV) algorithm allowing us to obtain Eulerian and Lagrangian statistics of the kinematic quantities as well as estimates of the snowflake characteristics like size, aspect ratio and orientation. In addition, by assuming a constant particle density, we derive particle mass flux profiles. The results show that the particle size distribution in saltation can indeed be characterized by a lognormal or a gamma distribution. From the analysis of the particle streamwise velocity profiles, it is clear that the assumption of a constant particle speed inside the saltation layer (common in simple saltation models) might not be a good approximation even for low friction velocities. We will present in how far we can assess the influence of the snow properties on mass flux and saltation dynamics as a basis to validate recent model results on the influence of inter-particle cohesion for example. Moreover, this data set will contribute to the development of new parameterizations for snow saltation mass flux and streamwise velocity that would take into account the effect of snow density and hardness.