A collision model reproduces Arctic sea-ice motion

Wind pushes Arctic sea ice, but by itself it does not explain how ice floes spread or the variety of observed speeds. A study published in Physical Review Letters shows that a model adding collisions among those floes reproduces patterns measured in Fram Strait, between Greenland and the Svalbard archipelago.

Sea ice is not a single plate. It breaks into floes ranging from meters to kilometers across, moving under the influence of wind and ocean water. When many floes are concentrated in one area, a floe encounters a neighbor before the wind direction changes substantially. Each collision transfers and dissipates part of the incoming energy.

Bryan Shaddy, P. Alex Greaney and Bhargav Rallabandi represented the ice as a granular material: many colliding objects exposed to variable winds and water drag. The model used local measurements of wind and ice conditions as inputs instead of freely tuning a long list of parameters to imitate the outcome.

The researchers compared the simulation with observations from Fram Strait. The same model reproduced floe dispersion, the distribution of their velocities and the motion's power spectrum — a measure of how variation is distributed across timescales from hours to days. Agreement across all three tests is more informative than matching only an average speed.

The proposed mechanism is simple. Collisions shorten the distance each floe travels freely and quickly remove part of the energy supplied by the wind. A kinetic theory based on the Boltzmann equation, which statistically describes many interacting particles, recovered patterns seen in both simulations and observations. It provides a bridge between small-scale impacts and ice transport at climate-relevant scales.

A preliminary version of the work had circulated since February 2025; the current development is the peer-reviewed publication on September 10, 2026, followed by the institutional release on September 16. The study is not a forecast of where ice will travel under future warming. It supplies a physical relationship that can be tested and incorporated into models unable to follow millions of floes individually.

The result therefore does not replace wind, currents, fractures or waves with a single cause. It shows that collisions combined with measured local conditions are sufficient to reproduce three previously difficult-to-reconcile statistics in Fram Strait. The next test is whether the same description works across other regions, ice concentrations and wind regimes.