Towards a fully unstructured ocean model for ice shelf cavity environments: Model development and verification using the Firedrake finite element framework

William Scott, Stephan C. Kramer, Paul R. Holland, Keith W. Nicholls, Martín J. Siegert, Matthew D. PiggottOcean 2023 · DOI

TL;DR

This paper introduces a new, flexible ocean model built with the Firedrake finite element framework to better simulate the complex environments beneath Antarctic ice shelves. This model, capable of handling irregular geometries, shows good agreement with established models and demonstrates promising capabilities for future research into ice shelf stability and sea-level rise.

What Problem Does This Paper Address?

Predicting how much ice Antarctica will lose to the ocean is critical for understanding future sea-level rise. A key factor is the melting that occurs beneath ice shelves, particularly near the "grounding zone" where ice transitions from resting on bedrock to floating. However, these regions are remote and difficult to study directly. Existing ocean models often use structured grids, which struggle to accurately represent the complex, irregular shapes of ice shelf cavities and the critical grounding zone. This limits our ability to accurately model melt rates and, consequently, to predict ice loss.

Methods

The authors developed a new nonhydrostatic ocean model using the Firedrake finite element framework. Firedrake allows for the use of unstructured meshes, meaning the model grid can conform to complex geometries without being constrained by rigid structures. This is a significant advantage for simulating the intricate shapes of ice shelf cavities. The model solves the full incompressible Navier-Stokes equations, including nonhydrostatic terms, which are important for capturing vertical dynamics in areas where horizontal and vertical scales are comparable. A key feature of Firedrake is its ability to automatically generate adjoint models, which are crucial for sensitivity analysis and optimization. The paper also details the finite element discretisation choices for velocity, pressure, temperature, and salinity, as well as the implementation of a standard "three-equation" melt parameterisation at the ice-ocean boundary.

Datasets

The paper uses the ISOMIP+ Oceano test case for evaluating the model's performance in a full ice shelf cavity simulation. This is a community standard benchmark designed for comparing different ocean models. Additionally, a Method of Manufactured Solutions (MMS) test was employed for rigorous verification of the model's numerical accuracy. No new datasets were introduced.

Key Insights and Ideas

  • Flexibility is Key: Unstructured meshes are essential for accurately representing the complex geometries of ice shelf cavities, especially near grounding zones and basal features like crevasses.
  • Nonhydrostatic Dynamics Matter: Forcing the model to include nonhydrostatic vertical acceleration terms is important for capturing the full physics in these environments, where aspect ratios can approach one.
  • Firedrake's Power: The Firedrake framework significantly streamlines the development of complex finite element models, particularly by enabling automatic adjoint code generation.
  • Melt Rate Sensitivity: Melt rates are highly sensitive to grid resolution, especially vertical resolution in traditional z-layer models. This highlights the need for careful verification and understanding of numerical mixing.
  • Adjoint Models as Powerful Tools: The automatically generated adjoint model provides valuable insights into how different physical parameters influence melt rates, offering a complementary approach to forward modeling for understanding complex processes.

Key Results

  • Verification: Method of Manufactured Solutions (MMS) tests confirmed that the model's discretisation is sound and achieves second-order accuracy for velocity, temperature, and melt rate.
  • ISOMIP+ Comparison: The new model, when run on the ISOMIP+ Oceano test case, produced results in good agreement with the widely used MITgcm model, particularly at higher vertical resolutions.
  • Melt Rate Behavior: The study confirmed that melt rates are sensitive to grid resolution, with the new unstructured mesh model showing less extreme dependence compared to MITgcm at coarser resolutions.
  • Adjoint Sensitivities: Preliminary adjoint calculations revealed that melt rate is sensitive to temperature and salinity, with buoyancy playing a significant role. Sensitivities were also concentrated near the grounding line and ice-ocean interface for viscosity and diffusivity.

Limitations and Future Work

  • Melt Parameterisation: The authors acknowledge that the current melt parameterisation, while standard, may not fully capture the complex physics of the ice-ocean boundary layer, particularly concerning turbulence and stratification. They suggest that future work could involve developing more robust parameterisations.
  • Turbulence Closure: The model currently uses a simple eddy diffusivity closure. Future work will involve implementing more advanced turbulence closure schemes (e.g., RANS, LES) to better represent mixing processes.
  • Computational Cost: While not the primary focus, the authors note that finite element methods with implicit time-stepping can be computationally more expensive than explicit finite volume methods. However, the ability to take longer timesteps and represent complex geometries can offset this.
  • Model Validation: Direct observations of sub-ice shelf conditions are still scarce, making model validation a persistent challenge.
  • Future Directions: The authors plan to apply the model to more complex geometries, investigate the impact of basal crevasses, and extend its adjoint capabilities for parameter estimation and data assimilation.

Summary updated

Abstract

Numerical studies of ice flow have consistently identified the grounding zone of outlet glaciers and ice streams (the region where ice starts to float) as crucial for predicting the rate of grounded ice loss to the ocean. Owing to the extreme environments and difficulty of access to ocean cavities beneath ice shelves, field observations are rare. Estimates of melt rates derived from satellites are also difficult to make near grounding zones with confidence. Therefore, numerical ocean models are important tools to investigate these critical and remote regions. The relative inflexibility of structured grid models means, however, that they can struggle to resolve these processes in irregular cavity geometries near grounding zones. To help solve this issue, we present a new nonhydrostatic unstructured mesh model for flow under ice shelves built using the Firedrake finite element framework. We demonstrate our ability to simulate full ice shelf cavity domains using the community standard ISOMIP+ Ocean0 test case and compare our results against those obtained with the popular MITgcm model. Good agreement is found between the two models, despite their use of different discretisation schemes and the sensitivity of the melt rate parameterisation to grid resolution. Verification tests based on the Method of Manufactured Solutions (MMS) show that the new model discretisation is sound and second-order accurate. A main driver behind using Firedrake is the availability of an automatically generated adjoint model. Our first adjoint calculations, of sensitivities of melt rate with respect to different inputs in an idealised grounding zone domain, are promising and point to the ability to address a number of important questions on ocean influence on ice shelf vulnerability in the future.
BibTeX
@inproceedings{scott2023towards,
  author = {William Scott and Stephan C. Kramer and Paul R. Holland and Keith W. Nicholls and Martín J. Siegert and Matthew D. Piggott},
  title = {{Towards a fully unstructured ocean model for ice shelf cavity environments: Model development and verification using the Firedrake finite element framework}},
  booktitle = {Ocean},
  year = {2023},
  doi = {10.1016/j.ocemod.2023.102178},
}
Figure