Abstract
Laterally propagating, through-cutting rifts are common features of Antarctic ice shelves. Over most of their advective histories, very little appears to happen to ice shelf rifts. Lateral propagation events are rare and rift tips are observed to "arrest" at structural boundaries within the floating ice. Typical linear elastic fracture mechanics (LEFM) analyses lead to two, possibly contradictory, conclusions about this: (1) lateral propagation would continue were it not for unspecified, heterogeneous ice properties and (2) rift opening is insufficiently supported by the ocean and there should be no lateral propagation at all. Nevertheless, rifts persist and do sometimes grow. Here, we show how melange (a material jumble of shelf-ice, sea-ice and firn) can facilitate lateral propagation where ocean-filling alone cannot, and consider why, apart from its inability to provide realistic lateral limits, the traditional LEFM works so well.
Rifts and the stress fields they embed are interdependent. Rifts are discontinuities in the ice shelf continuum that disrupt stress transmission and introduce new internal boundaries. Both viscous and brittle deformation of the ice are a response to the stress field, and as those responses progress, the stress field is modified. In this work, we present a framework for explicit representation of rifts in an ice shelf model that by its nature represents this interdependence. We apply LEFM with an extended finite element method (XFEM) scheme to discretize a 2D model of the ice shelf with an included rift. Propagation is predicted using stress intensity factors (SIFs) calculated near the tips. The 2D view of the shelf for elasticity is established to be consistent with the shallow-shelf-approximation and a 3D model is used to demonstrate a consistent (reproducible) depth dependence of the SIFs, further justifying the 2D approach. We apply a suite of boundary conditions on a realistic mid-shelf rift to consider three possibilities: (1) no control in a null-support case, (2) over-control in the ocean-support (ocean-and-air-filled) case and (3) limited control in an ocean-and-melange support case, in which propagation occurs at one extreme of melange behaviour and does not at the other. TLDR; melange causes ice shelf rifts to propagate, in a limited way. The code is available on github, ymmv.