Non-Gaussian statistics of inelastically interacting ice floes

Sea ice comprises plates (or floes) of ice that are transported on the ocean surface by wind and ocean currents. Despite the typically Gaussian noise of the driving wind, observations show a much longer tailed velocity distribution of the ice, which is not well understood. Here, we present a stochastic dynamics framework of interacting ice floes that resolves the connection between wind and ice statistics. We model the wind as the superposition of a mean and a normally distributed single-correlation-time noise. The noisy wind drives a dynamical equation for the ice floes that includes ice inertia, ocean drag, and inelastic collisions between the floes. Under this framework, we find that the ice velocity distribution is indeed broader than that of the driving wind. This broadening is controlled by collisions, which depend on ice inertia, floe size and area fraction, as well as the wind statistics. Our results for the ice velocity distribution and frequency spectrum are in good agreement with observations in the Fram Strait, without the need to invoke details of the wind turbulence. We rationalize our findings in terms of a coarse-grained Fokker-Planck framework, which can ultimately inform climate-scale parametrizations based on floe-scale physical principles.