The increasing catalogue of high-quality ice-penetrating radar data provides a unique insight in the internal layering architecture of the Greenland ice sheet. The stratigraphy, an indicator of past deformation, highlights irregularities in ice flow and reveals large perturbations without obvious links to bedrock shape. In this work, to establish a new conceptual model for the formation process, we analysed the radar data at the onset of the Petermann Glacier, North Greenland, and created a three-dimensional model of several distinct stratigraphic layers. We demonstrate that the dominant structures are cylindrical folds sub-parallel to the ice flow. By numerical modelling, we show that these folds can be formed by lateral compression of mechanically anisotropic ice, while a general viscosity contrast between layers would not lead to folding for the same boundary conditions. We conclude that the folds primarily form by converging flow as the mechanically anisotropic ice is channelled towards the glacier.

In times of warming in polar regions, the prediction of ice sheet discharge is of utmost importance to society, because of its impact on sea level rise. In simulations the flow rate of ice is usually implemented as proportional to the differential stress to the power of the exponentn = 3. This exponent influences the softness of the modeled ice, as higher values would produce faster flow under equal stress. We show that the stress exponent, which best fits the observed state of the Greenland Ice Sheet, equalsn = 4. Our results, which are not dependent on a possible basal sliding component of flow, indicate that most of the interior northern ice sheet is currently frozen to bedrock, except for the large ice streams and marginal ice.

Synthetic polycrystalline ice was sheared at temperatures of −5, −20 and −30 ^∘C, to different shear strains, up to γ=2.6, equivalent to a maximum stretch of 2.94 (final line length is 2.94 times the original length). Cryo-electron backscatter diffraction (EBSD) analysis shows that basal intracrystalline slip planes become preferentially oriented parallel to the shear plane in all experiments, with a primary cluster of crystal c axes (the c axis is perpendicular to the basal plane) perpendicular to the shear plane. In all except the two highest-strain experiments at −30 ^∘C, a secondary cluster of c axes is observed, at an angle to the primary cluster. With increasing strain, the primary c-axis cluster strengthens. With increasing temperature, both clusters strengthen. In the −5 ^∘C experiments, the angle between the two clusters reduces with strain. The c-axis clusters are elongated perpendicular to the shear direction. This elongation increases with increasing shear strain and with decreasing temperature. Highly curved grain boundaries are more prevalent in samples sheared at higher temperatures. At each temperature, the proportion of curved boundaries decreases with increasing shear strain. Subgrains are observed in all samples. Microstructural interpretations and comparisons of the data from experimentally sheared samples with numerical models suggest that the observed crystallographic orientation patterns result from a balance of the rates of lattice rotation (during dislocation creep) and growth of grains by strain-induced grain boundary migration (GBM). GBM is faster at higher temperatures and becomes less important as shear strain increases. These observations and interpretations provide a hypothesis to be tested in further experiments and using numerical models, with the ultimate goal of aiding the interpretation of crystallographic preferred orientations in naturally deformed ice.