Modeling the fractal geometry of arctic melt ponds using the level sets of random surfaces (Q1645014)

The melting of snow and sea ice in Arctic summer causes the formation of \textit{melt ponds} whose reflection properties differ from the ambient snow and ice. The formation process of these melt ponds changes the ice albedo and is a significant factor in current clima models. Thus, an accurate and yet computationally tractable mathematical description of melt pond formation during Arctic summer is of importance. Aearial photo series suggest a fractal boundary of melt ponds and a transition in fractal dimension from 1 to about 2. The authors suggest a simple melt pond model obtained by intersecting a random surface with a horizontal plane whose height increases with time. The construction of a suitable random surface is the mathematical core of this article. The Ansatz is a two-dimensional Fourier series with random coefficients that are determined in order to fit data drawn from aerial photos. A more sophisticated model also takes into account the melting process itself via a partial differential equation. The authors evaluate the influence of random coefficients in numerical simulations for both models and discuss their applicability in climate models. The find some differences in fractal dimension shift but yet consider both models to be sufficiently realistic. The simpler non-PDE model is assessed as a computationally efficient and conceptually simple way to capture first order effects.