Disintegration of Marine Ice-sheets Using Novel Optimised Simulations

The Thwaites Glacier is currently unstable and vulnerable to future collapse. The DOMINOS (Disintegration of Marine Ice-sheets Using Novel Optimised Simulations) project aims to reduce uncertainty in the rate of mass loss from this influential glacier and adjacent regions of the West Antarctic Ice Sheet. Specifically, the project will:

  • use a suite of ice and ocean models to explicitly simulate iceberg calving and associated dynamic processes,
  • employ this model suite to improve our understanding of how these processes interplay with climatic and oceanic forcing; and
  • inform how to accurately represent these processes in ice sheet models for larger-scale, longer-term projections of ice shelf and glacier retreat.

The results of DOMINOS will contribute to the International Thwaites Glacier Collaboration, a NERC-NSF research initiative that is investigating the stability and retreat of this large Antarctic outlet glacier to better constrain predictions of future sea level rise (SLR) and inform international climate change adaptation policy.

Connections between International Thwaites Glacier Collaboration projects and their external impact pathways. Modified from original image by D. Vaughn.
Thwaites Glacier as imaged by Sentinel-1 synthetic aperture radar on 10 March 2019. (c) EC Copernicus data/ESA/CMEMS/Polar View


Thwaites Glacier, like other outlet glaciers that are situated on retrograde slopes, is projected to retreat rapidly due to feedback processes linked to oceanic and atmospheric forcing. This rapid retreat will have significant consequences for SLR, as Thwaites Glacier drains a large portion (192,000 km2) of the West Antarctic Ice Sheet. Providing more reliable predictions of the future behavior of the glacier requires the use of novel modelling techniques that accurately simulate calving and associated dynamic processes under changing climate scenarios. This modelling, in turn, necessitates a detailed understanding of the fundamental processes that control calving.

The processes and feedbacks of marine ice sheet instability and marine ice cliff instability. From De Conto and Pollard (2016)


Marine ice sheet instability (MISI) occurs when the grounding line of marinate-terminating glaciers is forced to retreat over a retrograde bed, which causes increasing ice discharge across the grounding line. Observations and modelling studies indicate that the process of MISI has already begun at Thwaites Glacier. The future retreat of Thwaites Glacier may also be exacerbated by marine ice cliff instability (MICI), a process in which an ice front that is rapidly retreating into a deepening basin becomes increasingly unstable. This could lead to runaway ice cliff failure and further ice sheet disintegration. DOMINOS will develop detailed models of ice-shelf processes and ice-cliff instability to identify and quantify the factors that control calving rates and stability of Thwaites Glacier.

Project Approach and Objectives

We will use a suite of ice and ocean models with overlapping spatial and temporal scales to: a) resolve the spectrum of processes that control calving, and b) investigate the interplay between these processes and atmospheric and oceanic forcing. The model suite includes the Helsinki Discrete Element Model (HiDEM), the 3D full Stokes continuum model Elmer/Ice, and the Berkeley Ice Sheet Initiative for Climate Extremes (BISICLES) model. Our ocean forcing model suite includes simple plume models, intermediate complexity 2-layer ocean models and the fully 3D regional model MIT General Circulation Model (MITgcm). We will use this model suite to: a) conduct explicit simulations of the processes implicated in MISI, b) develop reliable parameterizations of these processes for predictive models, and c) simulate the full dynamic response of Thwaites Glacier to climate change on decadal to centennial time scales. We will meet the following interrelated set of objectives to achieve this.

O1: Develop detailed models of ice-shelf processes and ice-cliff instability to identify and quantify the factors that control calving rates and stability of Thwaites Glacier

Ice shelf calving, disintegration and MICI processes will be modelled using HiDEM and Elmer/Ice, coupled to submarine- and surface-melt models. Discrete element models, such as HiDEM, represent materials as arrays of particles connected by breakable elastic beams and can resolve fracture and iceberg calving events. HiDEM will be combined with ‘Elmer/Ice’, which simulates stress and velocity fields across a glacier. Model experiments will focus on identifying system thresholds and rate-controlling processes under a wide range of boundary conditions.

Example output of an Elmer/Ice simulation of calving and associated stress fields at Store Glacier. Animation courtesy of J. Todd.

A test HiDEM simulation of the evolution of an idealized Thwaites Glacier. Animation courtesy of J. Astrom.

O2: Parameterize ice-shelf and ice-cliff instability processes for use in predictive ice sheet modelsHiDEM and Elmer/Ice have exceptionally high fidelity but are too computationally expensive to perform over large areas for long duration runs. It is therefore necessary to employ dynamic models with simpler stress formulations and representations of calving processes to simulate ice sheet dynamics on these longer timescales. We will use the results of the high-fidelity model experiments to develop suitable parameterizations of calving and related dynamic processes for the BISICLES ice sheet model. We will run BISICLES alongside select HiDEM and Elmer/Ice simulations to ensure that these processes are realistically represented in the larger-scale ice sheet model. For additional validation, the performance of all three models will be tested using observed terminus positions and retreat rates for select locations in Antarctica and Greenland.

O3: Predict the response of Thwaites Glacier to different forcing scenarios over decadal and century timescales to examine the sensitivity of the response to alternative forcings.The optimized BISICLES ice sheet model will be used to simulate the evolution of Thwaites Glacier under a range of forcings derived from CMIP6 (Coupled Model Intercomparison Project) projections. Model runs will span decades to centuries. As Thwaites Glacier is not dynamically isolated from nearby glacier catchment basins (e.g. Pine Island Glacier), longer runs will encompass the entire Amundsen Sea Embayment to allow for migration of these catchment boundaries. If needed, a set of limited simulations will encompass all of the West Antarctic Ice Sheet. Simulations will employ a range of model parameters and forcing scenarios to bracket rates of mass loss and associated SLR.

Project Team

Doug Benn (PI) is Professor of Environmental Change at the University of St Andrews. His 30 year experience of glaciological research includes the response of glaciers to climate change, glacio-speleology, glacier surges, iceberg calving and glacier dynamics.

Tom Cowton (Co-I) is lecturer in Physical Geography at the University of St Andrews, with expertise in numerical modelling of feedbacks between glacial runoff, submarine melting and near-glacier ocean circulation using MITgcm and plume theory.

Anna Crawford (PDRF) is a postdoctoral research fellow at the University of St Andrews. Her previous research focused on the deterioration of ice islands that originate in northern Greenland, from which she gained knowledge and experience in ice-ocean interactions, numerical modelling, 3D model generation, and field data collection.

Project Partners

Adrian Luckman is Professor of Glaciology and Remote Sensing at the University of Swansea. He has 20 years of experience of satellite remote sensing of the cryosphere, including calving processes, ice dynamics using feature tracking on Synthetic Aperture radar data, and ice shelf processes and stability.

Project Collaborators

This project will continue the close collaboration between the St Andrews Glaciology Group and Jan Åström and Thomas Zwinger at the IT Center for Science, Helsinki. Jan is the developer of the Helsinki Discrete Element Model (HiDEM) and Thomas is a core developer of Elmer/Ice.