27 January, 2010 (16:00 GMT), 5 Merrion Square, Dublin 2.

Speaker: Ondrej Soucek, DIAS.
Title: Numerical modeling of ice sheet dynamics.

Abstract:

Ice sheets are masses of grounded ice, which evolve in regions where climatic conditions allow for long-term deposition of snow cover. Through a complex process of successive compaction inducing internal structure changes, glacier ice is formed, being one of the many known crystalline phases of solid water. At sufficiently long time scales, due to internal creep, glacier ice behaves as a non-Newtonian fluid. This fact enables us to describe and explain the evolution of ice-sheets at these time scales by means of continuum thermomechanics constructing a nonlinear-fluid model with geometry that is controlled by gravitationally driven creep flow that is coupled with thermal conditions inside the ice sheet, and by surface processes of accumulation, ablation and basal sliding. Since ice-sheets are typically flat, with the vertical-to-horizontal aspect ratio smaller than 1/100, a scaling approximation utilizing this fact is often adopted in the glaciological community, resulting in the so-called “shallow-ice” approximation (SIA). This approximation of ice flow enables to compute the ice-sheet velocity field, induced by gravity, semi-analytically, which represents an effective computational tool compared to more accurate approaches. During the last several years, however, the shallow-ice approximation is being slowly abandoned, as the effects of higher-order dynamics or even the exact solution to the ice-flow problem are looked for, typically, by means of advanced numerical techniques such as finiteelements or spectral methods. The increase of computational demands is, however, enormous compared to the SIA, making it problematic to implement these techniques for the large-scale evolutionary ice-sheet models. We therefore designed an iterative algorithm, capable of successive improvement of the SIA solution towards the exact solution, still possessing the SIA’s computational effectiveness. To demonstrate the applicability of our technique, we will present results from (i) the Ice-Sheet Model Intercomparison Project – Higher-Order Models (ISMIP-HOM), a benchmark focused on the evaluation of the non-shallow higher-order effects the on ice-sheet dynamics, (ii) from
the European Ice Sheet Modeling INiTiative (EISMINT) benchmark focused on investigating the large-scale characteristics of thermo-mechanically coupled ice-sheet models and (iii) from the EISMINT benchmark for Greenland Ice Sheet models.