3D Snowfall

Flakes fall at the speed set by drag balance for their crystal type, drift in a turbulent wind and pile up on a measuring board in time-lapse.

Switch to 2D

Presets

Playback

Weather

Wind

View

Camera

Layers

Quality

Model check

The science

A snowflake reaches terminal velocity when its weight equals the aerodynamic drag: m g = ½ ρa Cd(Re) A vt², with Re = ρa vt L/η. For a sphere Cd = 24/Re (1 + 0.15 Re0.687) + 0.42/(1 + 4.25×104 Re-1.16) (Clift-Gauvin); at Re → 0 it becomes Stokes' law v = 2r²(ρi - ρa) g/(9η). Air density follows the ideal gas law and viscosity follows Sutherland's law, so cold air is denser and thicker.

The crystal habit comes from the Nakaya diagram: plates near 0 to -4 °C, needles around -5, hollow columns to -10, plates and then stellar dendrites near -15, plates again, and columns below -22 °C. Each habit has a representative mass and projected area (a plate of thickness 0.0138 D0.474 mm, a dendrite with 35 % fill, needles and columns with width growing as a power of length, aggregates with effective density 70 D-1.1 kg/m³, graupel at 100-400 kg/m³). Riming thickens crystals and raises graupel density, so rimed particles fall faster.

The flake number flux is the water flux divided by the flake mass, Φ = R/m, and the number concentration is n = Φ/vt. The extinction coefficient β = 2 n A (geometric optics) gives the Koschmieder visibility V = 3.912/β. Fresh-snow density uses ρs = 67.9 + 51.3 eT/2.59 (Hedstrom and Pomeroy 1998), so the depth rate is R ρw/ρs: dry powder at -12 °C piles up about 14 times deeper than the water it contains, wet snow near 0 °C only 8 times.

Simplifications, stated honestly. Flakes are drawn as one habit and one size at a time (real snowfall is a size distribution), their on-screen size is exaggerated by the display slider, and the drawn count is a sample of the real number concentration. Wind is a mean flow plus a synthetic sum-of-sinusoids turbulence, not a solved flow, flakes do not rotate or collide, and drifting, compaction, melting and sublimation of the snowpack are ignored. The cabin roof catches snow but no wind shadow forms behind it. The shape factors and mass laws are representative fits, so speeds are right to roughly a factor of 1.5, not exact.

Loading model...
Loading Blender model...
Terminal velocity against flake size for every crystal habit (white dot: the flake now)
Snow depth on the measuring board against time-lapse time
Visibility against precipitation rate for the current flake type
Made with 🟠 Blender · Three.js