Inferred basal friction and mass flux affected by crystal-orientation fabrics

Abstract

We investigate the errors caused by neglecting the crystal-orientation fabric when inferring the basal friction coefficient field, and whether such errors can be alleviated by inferring an isotropic enhancement factor field to compensate for missing fabric information. We calculate the steady states that arise from ice flowing over a sticky spot and a bedrock bump using a vertical-slab numerical ice-flow model, consisting of a Weertman sliding law and the anisotropic Johnson flow law, coupled to a spectral fabric model of lattice rotation and dynamic recrystallisation. Given the steady or transient states as input for a canonical adjoint-based inversion, we find that Glen's isotropic flow law cannot necessarily be used to infer the true basal drag or friction coefficient field, which are obscured by the orientation fabric, thus potentially affecting vertically integrated mass fluxes. By inverting for an equivalent isotropic enhancement factor, a more accurate mass flux can be recovered, suggesting that joint inversions for basal friction and the isotropic flow-rate factor may be able to compensate for mechanical anisotropies caused by the fabric. Thus, in addition to other sources of rheological uncertainty, fabric might complicate attempts to relate subglacial conditions to basal properties inferred from an inversion relying on Glen's law.

Crystallographic Preferred Orientation (CPO) Development Governs Strain Weakening in Ice: Insights From High‐Temperature Deformation Experiments

“…This weakening is often referred to as strain rate “enhancement” in the glaciological and ice sheet literature (Alley, 1992; Budd & Jacka, 1989; Treverrow et al., 2012), and is often accounted for in large‐scale ice dynamics models (e.g., Azuma, 1995; Morland & Staroszczyk, 2009; Placidi et al., 2010). Notably, recent work has shown that enhancement must be accounted for in order to accurately infer basal friction—and therefore solid ice discharge—of a flowing ice mass (Rathmann & Lilien, 2021)…”

Accelerating ice flow at the onset of the Northeast Greenland Ice Stream

“…Modeling can help us distinguish between different expected responses to hypothesized forcing scenarios. We run four different experiments with an idealized model of a 2D cross section of an ice stream to investigate the surface velocity response to changes in the boundary conditions and ice-flow parameters (see online methods 4) 19 – 21 . Results are consistent with physical theory, but have greater spatial detail: (1) a thicker ice stream flows faster and but, unlike observations, the velocity anomaly is concentrated in the ice-stream interior; (2) increasing the zone of sliding at the base leads to a spatial map of acceleration that peaks just outside the shear margin; and (3) softening the ice column beneath the shear margin leads to an acceleration that peaks just inside the shear margin.…”

Crystal fabric anisotropy causes directional hardening of the Northeast Greenland Ice Stream

“…Strong COFs have long been known to change the directional viscosity of ice by several orders of magnitude compared to isotropic ice [41,42]. We evaluate the effect of the fabric-induced directional viscosities over NEGIS by calculating bulk directional enhancement factors [30,7] based on model-and radar-derived horizontal anisotropy. The enhancement factors are defined as the longitudinal and shear strain rates with respect to the principal fabric directions, divided by the expected strain-rate magnitude assuming isotropic ice.…”

“…Most state-of-the-art large-scale ice-flow models either ignore the mechanical anisotropy of ice entirely [6] or infer an isotropic enhancement factor that subsumes some effect of anisotropy [7,8].…”

“…Anisotropy may explain some of the major discrepancies between modelled and observed surface velocities in highly dynamic areas such as ice streams [9,10], since the commonly used isotropic flow law is no longer valid when COF is not isotropic. The resulting mechanical anisotropy potentially introduces errors in modelled strain rates, and thus bias in basal sliding velocities obtained by inversion methods [7].…”