First Direct Observations of Internal Flow Structures in a Powder Snow Avalanche: Turbulence, Instability and Particle Distribution

This study presents the first direct high-speed optical observations of individual particle motion within a natural powder snow avalanche, revealing distinct flow phases, quantifying turbulence and shear instabilities, and providing critical empirical data to refine numerical models of multiphase gravity currents.

The Gist

Imagine a massive, roaring cloud of snow and air tearing down a mountain. To the naked eye, it looks like a single, chaotic white blur. But for the first time, scientists have put on "super-vision" goggles to see what's actually happening inside that cloud.

This paper is about a groundbreaking experiment where researchers used high-speed cameras to watch individual snowflakes dance inside a real, natural powder snow avalanche. Here is the story of what they found, explained simply.

The Setup: A Snowy Laboratory

The researchers set up a "speed trap" on a mountain in Switzerland. They didn't just use radar (which sees the whole cloud like a blurry silhouette); they installed three high-speed cameras on a tall metal tower, right in the path of the avalanche.

Think of these cameras as super-fast eyes, snapping 1,000 pictures every second. They also used a giant LED light to shine a thin sheet of light through the air, illuminating the snow particles so the cameras could see exactly how they moved, how fast they were going, and how they clumped together.

The Three Acts of the Avalanche

The scientists realized the avalanche wasn't just one big mess. It was like a play with three distinct acts, each with its own personality:

Act 1: The "Sprinter" (The Surge)
At the very front, a fast, short burst of snow shot past the cameras. It was like a sprinter exploding out of the starting blocks.

• What happened: This was a dense, fast-moving wave of snow that didn't last long. It was chaotic, with big clumps of snow flying together and empty spaces in between.
• The feeling: Imagine a sudden gust of wind that knocks over a stack of books. It's fast, violent, and over quickly.

Act 2: The "Dance Party" (The Suspension)
Behind the sprinter came the main body of the avalanche. This is the part we usually think of as the "cloud."

• What happened: This was the most exciting part. The snow wasn't just falling; it was being held up by a wild, churning storm of air. The researchers saw giant swirling eddies (like giant whirlpools in the air) that were huge—some as big as a two-story house!
• The Analogy: Imagine a blender full of water and fruit. The blades spin, creating huge swirls that keep the fruit suspended. In the avalanche, the "blades" were the wind shear (air moving at different speeds), creating giant waves that kept the snow floating. They even saw Kelvin-Helmholtz instabilities—fancy scientific terms for the "curling waves" you see when wind blows over water, or when you peel back a layer of paint. These waves were mixing the snow and air violently.

Act 3: The "Settling" (The Wake)
Finally, the energy ran out. The "dance party" ended.

• What happened: The wild swirling stopped. The snow particles, no longer held up by the crazy turbulence, began to gently drift down to the ground, like snowflakes falling on a quiet winter day.
• The feeling: The chaotic blender turned off, and the fruit slowly sank to the bottom of the glass.

The Big Discoveries

1. The Snow Wasn't "Mixed"
Old models thought the snow cloud was like a smooth, well-mixed milkshake. The cameras proved this wrong. The snow was actually clumpy. It formed giant patches of dense snow separated by empty air, like a cloud of cotton candy that's been torn apart. This "clumping" is crucial because it changes how much force the avalanche hits the ground with.

2. The Snow Particles Have "Muscle"
The snow particles were surprisingly heavy and big (about the size of a pea or a small marble). Because they were heavy, they didn't just follow the air perfectly. They had their own momentum.

• The Analogy: Imagine throwing a bowling ball and a feather into a strong wind. The feather follows the wind perfectly. The bowling ball fights the wind and keeps going straight. The snow in this avalanche was more like the bowling ball. It was too heavy to be a perfect "tracer" of the air, which makes predicting its path much harder.

3. The "Rolling Waves" Drive the Chaos
The researchers confirmed that the reason the snow stayed in the air for so long wasn't just random wind. It was driven by those giant, rolling waves (instabilities) they spotted. These waves act like a conveyor belt, constantly lifting the snow up and keeping it suspended. Without these waves, the snow would have fallen to the ground much sooner.

Why Does This Matter?

For years, scientists have tried to predict avalanches using computer models. But those models were like trying to predict traffic by only looking at the map, without knowing how individual cars brake or swerve. They were missing the "particle-scale" details.

This study provides the first real-life "footage" of how snow behaves inside the cloud.

• Better Safety: By understanding that the snow is clumpy and driven by giant waves, engineers can build better models to predict exactly how far an avalanche will travel and how hard it will hit buildings.
• Universal Physics: The same physics that governs a snow avalanche also governs underwater mudslides (turbidity currents) and volcanic ash clouds (pyroclastic flows). By solving the puzzle of the snow avalanche, we are also learning how to predict these other deadly natural disasters.

In short: This paper took the "blur" out of the avalanche. It showed us that a powder snow avalanche isn't just a falling cloud; it's a complex, churning, wave-driven engine that keeps snow suspended in the air, only to let it settle when the engine finally runs out of fuel.

Technical Summary

1. Problem Statement

Powder snow avalanches (PSAs) are destructive, multiphase gravity currents consisting of a dense basal layer and an airborne suspension layer where snow particles are suspended in turbulent air. Despite their hazard potential, the internal dynamics of the airborne layers remain poorly understood due to a fundamental lack of direct, high-resolution, particle-scale field data.

• Limitations of Previous Studies: Existing measurements (e.g., impact pressure sensors, ultrasonic anemometers, radar) provide macroscopic data but lack the spatial and temporal resolution to resolve individual particle motion, turbulence length scales, and shear instabilities within the dilute suspension.
• Knowledge Gaps: Key unresolved questions include the validation of the three-layer conceptual model (basal, transition, suspension), the specific characteristics of turbulence (integral scales, energy cascades), and the role of shear-driven instabilities (e.g., Kelvin–Helmholtz) in driving mixing and entrainment.

2. Methodology

The study utilized a multi-sensor approach at the Vallée de la Sionne (VdlS) full-scale avalanche test site in Switzerland to observe a naturally released avalanche (Event #20243024).

• Instrumentation:
  • GEODAR (GEOphysical flow dynamics using pulsed Doppler radAR): Provided macroscopic velocity and flow depth data from release to deposition.
  • Optical Sensors: Located at the base of a 20m pylon, measuring horizontal velocities in the dense basal layer via cross-correlation.
  • High-Speed Camera Array (Novel Component): Three high-speed cameras (1000 FPS, 1920x1080 px) mounted at heights of 5.5 m, 7.5 m, and 10.5 m on the pylon.
    • Illuminated by a planar LED light sheet (12° x 60°) perpendicular to the field of view to isolate particles.
    • Captured the transition and suspension layers with high spatial resolution.
• Data Processing:
  • Particle Image Velocimetry (PIV): Used to derive spatially and temporally resolved velocity fields from consecutive image pairs.
  • Image Analysis: Particle concentration was estimated via pixel intensity ratios; particle sizes were derived from best-fit ellipses on binarized images.
  • Turbulence Analysis: Integral length scales ($L_{int}$) were calculated from autocorrelation functions of velocity fluctuations. Continuous Wavelet Transform (CWT) was applied to identify multi-scale fluctuations and coherent structures.
  • Stability Analysis: Bulk Richardson numbers ($J$) and nondimensional wavenumbers ($\alpha_r$) were computed to assess shear instability conditions, comparing results against theoretical neutral stability curves.

3. Key Contributions

• First Direct Optical Observations: This study provides the first high-resolution, in-situ visualization of individual snow particle motion within the airborne layers of a natural PSA.
• Flow Segmentation: The authors successfully segmented the avalanche evolution into three distinct dynamic regions based on optical and velocity data:
  1. Region I (Early Surge): A short-lived, fast-moving, dilute front.
  2. Region II (Main Suspension): A highly dynamic phase with strong turbulence and particle clustering.
  3. Region III (Stabilized Wake): A decaying phase dominated by gravitational settling and stratification.
• Quantification of Instabilities: The study offers empirical evidence linking large-scale flow fluctuations to shear-driven instabilities (Kelvin–Helmholtz type), previously only hypothesized in PSAs.

4. Key Results

A. Flow Structure and Velocity Profiles

• Three-Region Evolution:
  • Region I: Characterized by a maximum velocity ($U_m$) of 14.2 m/s at a low height (2.4 m). The flow was shallow and highly inhomogeneous.
  • Region II: Velocity decreased ($U_m \approx 7-8$ m/s) but the flow depth expanded vertically, exceeding the 11 m measurement limit. The velocity maximum shifted upward as the suspension cloud grew.
  • Region III: Further deceleration ($U_m \approx 5.2$ m/s) with a broad vertical distribution, indicating a transition to a settling-dominated regime.
• Velocity Profiles: Profiles were fitted using a modified Power-Gaussian function incorporating a Navier slip condition. Significant basal slip velocities were observed, varying from ~10 m/s in the surge to ~2–3 m/s in the wake.

B. Turbulence Characteristics

• Integral Length Scales ($L_{int}$):
  • In Region I, $L_{int}$ was comparable to the flow depth (~5–6 m), suggesting depth-filling eddies.
  • In Region II, two distinct scales emerged: smaller eddies confined below the velocity maximum (shear with the bed) and larger eddies extending above it (shear with ambient air).
  • In Region III, scales oscillated widely, driven largely by ambient atmospheric turbulence rather than avalanche dynamics.
• Wavelet Analysis: Revealed coherent, long-period fluctuations (2–6 s) that exceeded the integral time scale, particularly in Region II. These were identified as organized motions rather than random turbulent cascades.

C. Instability Analysis

• Shear Instabilities: Stability analysis using the Bulk Richardson number ($J$) and wavenumber ($\alpha_r$) placed Region I and early Region II in or near the unstable regime.
• Kelvin–Helmholtz (KH) Billows: Visual evidence of wave-like structures at the flow interface, combined with stability analysis, confirms the presence of KH-type shear instabilities. These instabilities are identified as a primary driver for mixing, entrainment, and particle clustering.

D. Particle Dynamics

• Particle Size: Detected particles had geometric mean diameters of 2.7–2.9 mm (log-normal distribution), significantly larger than some previous estimates (0.16–0.2 mm) which likely sampled only the wake.
• Stokes Number ($St$):
  • $St_{int} < 1$: Particles respond to energy-containing turbulent eddies.
  • $St_{\eta} \gg 1$: Particles behave ballistically relative to fine-scale (Kolmogorov) fluctuations.
• Settling: In Regions I and II, turbulent suspension dominated ($V_t / \bar{u} < 1$). In Region III, settling became comparable to mean and fluctuating velocities, marking the transition to deposition.

5. Significance and Implications

• Model Refinement: The findings challenge the assumption of well-mixed, uniform suspension layers in current numerical models. The data suggests that Euler-Lagrange approaches (e.g., Large-Eddy Simulations with inertial particles) or advanced Euler-Euler closures are necessary to capture the observed heterogeneity and inertia effects.
• Hazard Assessment: The identification of shear instabilities and large-scale coherent structures explains the "pulsing" nature of avalanches and the generation of high-pressure pulses, which are critical for accurate hazard mapping and infrastructure design.
• Cross-Disciplinary Relevance: The study establishes strong analogies between PSAs, turbidity currents, and pyroclastic density currents, suggesting that the physics of shear-driven instabilities and particle-turbulence coupling are universal across particle-laden geophysical flows.
• Empirical Benchmark: The dataset provides the first empirical constraints for turbulence and instability dynamics in airborne avalanche layers, serving as a critical benchmark for validating future multiphase flow simulations.

In conclusion, this work bridges a critical gap between macroscopic observations and micro-scale physics in avalanche science, demonstrating that internal shear instabilities and particle inertia are fundamental drivers of avalanche evolution and destruction potential.