Speaker: Thomas Zwinger, CSC Helsingfors, Finland
Time: 7 February, 12.00-13.15
Plats: Ångström 11167

ABSTRACT

Sufficient long time-scales, ice can be treated as an incompressible, shear thinning fluid with a strong coupling of its material parameters to its temperature. As a consequence of the high viscosities, inertia is
negligible (vanishing Reynolds number) and the dynamics is governed by the Stokes equations. For decades, numerical models utilized certain approximations to these equations, commonly based on small values of the non-dimensional group defined by the ratio of characteristic horizontal to vertical spatial scales. As the shallowness assumption holds on a large scale of ice sheets (continental scale ice masses), ice shelves (floating ice nourished from inland ice), ice caps and even for most valley glaciers, certain situations occur, where it is violated. This is either due to dynamical aspects (e.g., large lateral or longitudinal gradients in velocities and/or stress components) or, simply, by geometry (strongly inclined surface, geometries with aspect ratios of unity order). Coincidently, these situations are linked to regions of particular interests for glaciologists, such as ice domes (dating of ice cores), ice streams and marine ice sheets (transition of ice sheets to shelves), the latter being of importance to quantify contributions of land-based ice masses to sea level rise. These aspects motivate the deployment of new generation ice flow models that are not limited by the shallowness assumption by solving  the unaltered Stokes equations. In particular, we present an open source, Finite Element Method (FEM) based suite, Elmer/Ice. Along a brief history of implemented problems using Elmer/Ice the possibilities, but also challenges imposed by the applied numerical methods are discussed.