Interferometric radar images collected by ERS-1, ERS-2 and RADARSAT-1 are used to observe the rupture tip of rifts that propagate along Hemmen Ice Rise on the Ronne Ice Shelf, Antarctica. Interferograms generated in 1992 and 1997 allow for the observation of ice deformation accumulated over 9 and 24 days respectively. These interferograms are combined, in order to separate the continuous process of creep deformation from the more cyclic motion caused by variations in ocean tide. An examination of local gradients in creep deformation reveals the pattern of ice deformation around and near the rupture tips and rifts with great precision (up to 10 cm a-1). We compare the observations with a deformation model for ice and obtain the following results: (1) The tidal oscillation of the Ronne Ice Shelf only yields small deformations along the rifts and near the rupture tips. (2) Along the ice front, the rifts and at the rupture tips, vertical bending is observed which is well explained by a model of viscous deformation of ice. Furthermore, the model indicates that the deformation pattern observed at the rupture tips is a sensitive indicator of the propagation state of the rifts (i.e. active vs inactive). (3) The viscous adjustment of ice is the dominant mode of deformation, masking the deformation pattern predicted by linear elastic fracture mechanics (LEFM). (4) Yet, at a spatial scale equivalent to the length of a rift, the propagation rate is well predicted by LEFM.

Ice flow numerical models are essential for predicting the evolution of ice sheets in a warming climate. Recent research emphasizes the need for higher-order and even full-Stokes flow models, instead of the traditional shallow-ice approximation, whose assumptions are not valid in certain critical areas. These higher-order models are, however, computationally intensive and difficult to use at the continental scale. Here we present a new technique, the Tiling method, to couple ice flow models of varying orders of complexity. The goal of the method is to limit the spatial extent of where higherorder models are applied to reduce the computational cost, while maintaining the model precision. We apply this method on synthetic geometries to demonstrate its practical use. We first use a geometry for which all models yield the same results to check the consistency of the method. Then we apply our method to a geometry for which a full-Stokes model is required in the vicinity of the ice front. Our results show that the hybrid models present significant improvements over mono-model approaches and reduce computational times.