Pick a scenario; every slider below stays editable.
The science
Atmosphere and updraft. Temperature falls 6.5 K per km (isothermal above 11 km); density follows the hydrostatic barometric law. The updraft is a Gaussian in height, w(z) = W exp(-((z - zp) / zw)²), times a Gaussian across the column, exp(-(x / Rc)²). Real updrafts are turbulent and vary in time; this is a steady idealisation.
Fall speed. Drag balances weight at the terminal speed vt = √(4 ρi g D / (3 Cd ρair)). For a 2 cm stone (ρi = 880 kg/m³, Cd = 0.6) that is about 20 m/s near 5 km. A stone is held aloft only where the updraft is at least vt, so the maximum hail size is set by the strongest updraft: Dmax = 3 Cd ρair W² / (4 ρi g). The stone's motion is integrated in two dimensions with RK4 using the full drag on the velocity relative to the moving air.
Growth by accretion. The stone sweeps up supercooled cloud droplets: dm/dt = E · LWC · A · |vrel| with cross-section A = πD²/4 and collection efficiency E. Cloud liquid water is zero above the -40 C homogeneous-freezing level and below the freezing level (where droplets are warm).
Dry versus wet growth (Schumann-Ludlam). Freezing releases latent heat Lf = 3.34e5 J/kg. The surface can lose heat only at Q = h π D² |Tair| with h = Nu k / D and Nu = 2 + 0.6 Re1/2 Pr1/3 (Ranz-Marshall). If the collected water exceeds what that heat budget can freeze, LWC > LWCcrit = 4 h |T| / (E v (Lf - cw|T|)), the surface stays wet (wet growth, clear ice) and the surplus is shed; below the limit everything freezes at once (dry growth, opaque rime). Real wet-growth stones can also retain water in a spongy layer; this model sheds it and ignores vapour diffusion.
Melting. Below the 0 C level the stone gains heat h π D² Tair and loses mass at Q / Lf (melt water is assumed shed). Small stones disappear completely before the ground; large ones survive with a smaller diameter. Evaporation, wet-bulb cooling and rain collision are ignored, so melting is somewhat over-estimated at the ground.
Honest simplifications. One stone in a steady idealised storm; no recirculation, no ice crystals or rain, no stone-stone collisions, no spin or irregular shape (a fixed Cd). The cutaway shells are painted from the model's own growth history: opaque shells are dry growth and clear shells are wet growth, as in real stones, but ring thicknesses are not data. Residence times here are longer than many observed values, and the liquid water profile is illustrative (peaks just above the freezing level and fades to zero at -40 C). Impact energy is the kinetic energy at ground level; damage thresholds depend on the surface and are not modelled. The 3D cloud, hills and houses are artistic and not to scale (houses and the stone are drawn larger than they are).