Modelling the three-dimensional, diagnostic anisotropy field of an ice rise

A. Clara J. Henry, Carlos Martín, Reinhard Drews

TL;DR

Researchers have developed a robust 3D model to simulate how ice crystals align—creating "fabric"—as they flow through an Antarctic ice rise. By testing different mathematical assumptions against radar-derived metrics, the study identifies where current observational assumptions fail and pinpoints the best locations for future field campaigns to improve ice-flow predictions.

What Problem Does This Paper Address?

While we often think of ice as a uniform, frozen block, at the scale of a glacier, it behaves like a highly complex fluid. Crucially, individual ice crystals are anisotropic: they deform much more easily along their "basal planes" (think of a slippery deck of cards) than in other directions. As ice flows, these crystals align into a preferred "fabric" that can significantly speed up or slow down the overall flow.

Despite its importance, most large-scale ice sheet models treat ice as an isotropic fluid (uniform in all directions) because modeling the 3D evolution of crystal orientation is computationally expensive and numerically unstable. Furthermore, there is a massive gap between the theoretical models and the data we get from the field (like radar or ice cores). This paper seeks to bridge that gap by simulating the 3D anisotropy field of the Derwael Ice Rise (DIR) in East Antarctica and providing a framework to compare these models directly with radar observations.

Methods

The authors utilized the Elmer/Ice finite element model to simulate the Derwael Ice Rise. Their approach involved several key technical choices:

  • One-Way Coupling: They used a fixed, steady-state 3D velocity field (from a previous study) to drive the evolution of the crystal orientation tensors ($a^{(2)}$ and $a^{(4)}$).
  • Semi-Lagrangian Numerical Method: To solve the evolution equations without the simulation "breaking" in high-stress areas, they employed a semi-Lagrangian scheme. This method is more stable than standard Eulerian approaches for advection-dominated problems.
  • Physical Constraints: They introduced a constraint ensuring the "degree of orientation" only increases over time, which prevents numerical dispersion from creating unphysical results.
  • Parameter Sensitivity: They tested four different combinations of two critical parameters: $\alpha$ (the relative influence of strain rate vs. deviatoric stress) and $\iota$ (the rate of fabric evolution). These parameters vary across existing glaciological literature, and this study highlights how much they actually matter.

Datasets

The study did not introduce a new raw dataset but synthesized several existing ones for the Derwael Ice Rise:

  • Input Velocity Field: A 3D steady-state velocity field generated by a thermomechanically-coupled full Stokes model (Henry et al., 2023).
  • Validation Context: The model results are designed to be compared with existing radar observations (Koch et al., 2023) and surface velocity maps (Rignot et al., 2017).
  • Benchmarking: The parameter combinations were pulled from four landmark studies (Martín et al. 2009, Seddik et al. 2011, Martín and Gudmundsson 2012, and Gagliardini et al. 2013).

Key Insights and Ideas

  • The "Vertical Assumption" is Risky: Radar processing often assumes that one of the crystal's principal axes is perfectly vertical. This paper shows that while this holds true in the center of grounded ice, it fails significantly in ice shelves and grounding zones, leading to errors in measured eigenvalues of up to 20%.
  • Strain Rate vs. Stress: The results suggest that the strain rate (how fast the ice is stretching) has a much stronger influence on developing a "single-maximum" fabric (where crystals point one way) at flow divides than the deviatoric stress does.
  • The "Tails" are Key: The areas of high horizontal divergence at the ends (tails) of the flow divide showed the most dramatic differences between the four model versions. This makes these "tails" the most valuable locations for future radar surveys to determine which physical parameters are actually correct.
  • Fabric Transitions: The model successfully captures the transition from isotropic ice at the surface to complex "girdle" fabrics (where crystals form an arc) and "single-maximum" fabrics at depth.

Key Results

  • Quantified Error: The study provides a concrete error estimate for radar observations: assuming a vertical eigenvector when the fabric is actually tilted can result in a significant underestimation of the smaller eigenvalues.
  • Fabric Mapping: In the ice rise interior, the largest eigenvalue ($\lambda_3$) is much larger than the others, indicating a strong single-maximum fabric. However, in the ice shelf, the fabric becomes more complex and varies based on whether the flow is dominated by extension or shearing.
  • Parameter Impact: The choice of $\alpha$ and $\iota$ isn't just a mathematical detail; it fundamentally changes the predicted fabric type in 3D space, particularly in complex flow regimes like grounding zones.

Limitations and Future Work

  • One-Way Coupling: The biggest limitation is that the fabric does not yet "talk back" to the flow. In reality, as the fabric develops, it changes the ice's viscosity, which in turn changes the velocity field. The authors acknowledge that "fully-coupled" 3D models are the ultimate goal but remain computationally daunting.
  • Steady-State Assumption: The model assumes the ice rise is in a steady state. While DIR is relatively stable, remnants of past flow regimes (transient effects) could still be "stored" in the deep ice.
  • Fabric Complexity: The current tensor approach cannot capture "cone" fabrics or multiple single maxima, which have been observed in some ice cores.
  • Next Steps: The authors recommend targeted radar campaigns in areas of high horizontal divergence to finally constrain the $\alpha$ and $\iota$ parameters, which would allow for more accurate global ice sheet modeling.

Summary updated

Abstract

Polar ice develops anisotropic crystal orientation fabrics under deformation, yet ice is most often modelled as an isotropic fluid. We present three-dimensional simulations of the crystal orientation fabric of Derwael Ice Rise including the surrounding ice shelf using a crystal orientation tensor evolution equation corresponding to a fixed velocity field. We use a semi-Lagrangian numerical method that constrains the degree of crystal orientation evolution to solve the equations in complex flow areas. We perform four simulations based on previous studies, altering the rate of evolution of the crystal anisotropy and its dependence on a combination of the strain rate and deviatoric stress tensors. We provide a framework for comparison with radar observations of the anisotropy field, outlining areas where the assumption of one vertical eigenvector may not hold and provide resulting errors in measured eigenvalues. We recognise the areas of high horizontal divergence at the ends of the flow divide as important areas to make comparisons with observations. Here, poorly constrained model parameters result in the largest difference in fabric type. These results are important in the planning of future campaigns for gathering data to constrain model parameters and as a link between observations and computationally-efficient, simplified models of anisotropy.
BibTeX
@article{henry2025modelling,
  author = {A. Clara J. Henry and Carlos Martín and Reinhard Drews},
  title = {{Modelling the three-dimensional, diagnostic anisotropy field of an ice rise}},
  journal = {Journal of Glaciology},
  year = {2025},
  doi = {10.1017/jog.2025.14},
}

Loading related content...

Figure