The biggest uncertainty in predicting global sea-level rise is what will happen to the polar ice sheets under different climate change scenarios. It is important for Aotearoa New Zealand to understand the impact of the Greenland Ice Sheet as well as Antarctic Ice Sheets, as gravitationally, the Southern Hemisphere is more affected by loss of ice from Greenland. Our team is at the forefront of various novel mathematical methods used to evaluate plausible scenarios for future polar ice sheet change, and these results will ultimately contribute to future global sea level projections.
Conduct statistical emulations of the Antarctic and Greenland Ice Sheets under warming scenarios
We will use new ice sheet model outputs (from the Antarctic Science Platform and Ice Sheet Model Intercomparison Project 6 & 7) to train a statistical emulator. The emulator will explore the sensitivity of the Antarctic Ice Sheet’s sea-level contribution to model parameters and to the IPCC’s range of shared socio-economic pathways. We will also apply the Antarctic Ice Sheet methodology to the Greenland Ice Sheet, which will improve our ability to assess global cryosphere contributions to sea-level rise.
Develop machine learning algorithms to characterize present-day changes in Antarctic Ice Sheet and Greenland Ice Sheet
We will build a set of process-focused (limited domain) simulations. Using these simulations, we will then develop and refine machine learning approaches that can accurately reproduce the way that each modelled ice sheet evolves. We will then undertake a suite of future focused scenario-based projections using the trained and validated machine-learning algorithm.
Develop machine learning algorithms to help refine existing parameterisations of ocean heat-driven ice shelf melt for future-forced simulations
We will build and train a machine learning framework that uses ocean temperature, salinity, and ice-shelf melt rate observations to produce optimized melt schemes for our process-based Ice Sheet Models. This approach reduces computational overhead and so enables many more scenarios to be explored. This in turn allows more robust projections of ice sheet mass loss than simply using process-based models with existing melt approximations. We will train our machine learning algorithms on high-resolution ocean models that capture eddy heat transfer across the continental shelf, developing parameterisations to inform and improve oceanographic representations in global-scale ocean models that may not currently accurately capture such fine-scale features.
Our Team
The leader of this research aim, Nick Golledge, is a world-class glaciologist and ice sheet modeller. He leads the National Antarctic Modelling Hub at Te Herenga Waka – Victoria University of Wellington, was a Lead Author on IPCC AR6, leads the Future Projections Expert Group (Antarctic Science Platform), and led Aotearoa’s contribution to WCRP ISMIP 6. He has published 10 papers in Nature & Science.
Dan Lowry and Mario Krapp, from Earth Science New Zealand (formally GNS Science), bring key numerical ice sheet modelling and data science expertise to the programme and are at the forefront of applying statistical emulation to evaluate model projection uncertainties. Contributions to this research is also being undertaken, from Te Herenga Waka – Victoria University of Wellington, by Stefan Jendersie, Bach Nguyen and Peter Siew.
Research
- The Ronne Ice Shelf Survived the Last Interglacial
- Modelling GNSS-Observed Seasonal Velocity Changes of the Ross Ice Shelf, Antarctica, Using the Ice-Sheet and Sea-Level System Model (ISSM)
- Insights on the Vulnerability of Antarctic Glaciers From the ISMIP6 Ice Sheet Model Ensemble and Associated Uncertainty
- Sensitivity of the Ross Ice Shelf to Environmental and Glaciological Controls