CONNECTing past, present and future: hindcast and forecast of Antarctic ice loss between 1990 and 2100

点击翻译按钮获取中文摘要

Satellite observations provide glaciologists with increasingly complete and frequent maps of ice velocity and thickness changes of the Antarctic Ice Sheet (AIS). For nearly three decades, satellites have helped us identify the physical processes that underlie contemporary rates of mass loss from the AIS, and have been key in quantifying Antarctica's present-day contribution to global sea level rise. However, the fundamental issue remains how these changes will evolve in a warming world, and what their global impact will be on sea level, climate and ecosystems. In order to provide reliable forecasts of ice sheet changes, and CONNECT the past, present and future, we need validated models with robust uncertainty quantification. This is what this project offers. In parallel with the satellite revolution, ice sheet models have advanced significantly over the last decade. Confidence in their ability to produce numerically robust projections for complex geometries and 'external' forcing has greatly improved, and essential dynamical interactions with the ocean and atmosphere can now be included. Yet, no systematic attempt has been made at comparing hindcasts of past changes from modern ice flow models with the rich, 30-year repository of satellite data from Antarctica. A model validation exercise of this type is both timely and critical: it needs the long time series of observations and sophisticated models that are now available, and it forms a prerequisite for reliable forecasts of the AIS's impact on global sea levels in the 21st century. In light of this important deficiency in ice sheet modelling and the need for robust, century-scale forecasts, De Rydt and his team of PDRAs and international project partners will address three distinct challenges: (1) the initial value problem of predicting the evolution of the Ice Sheet given an uncertain estimate of its present-day state, (2) the structural problem of unknown/uncertain physical parameters, and (3) the boundary value problem of assessing future changes in the state of the Ice Sheet due to uncertain future climate forcing.Building on De Rydt's internationally recognised expertise in ice dynamics and ice-ocean modelling, the team will first generate a best estimate of the state of the AIS in the 1990s through data assimilation in a coupled ice-ocean model, and perform the first circumpolar Bayesian uncertainty analysis to quantify how errors in the initial conditions are amplified by forward integration with realistic forcing between 1990 and 2020. Next, they will use newly developed techniques in perturbed physics ensembles in combination with model emulators to systematically sample uncertain model physics (basal sliding, ice rheology, calving, etc) and assess their impact on forecasts within a probabilistic framework. This will allow them to comprehensively validate a next-generation coupled ice-ocean model for the first time and identify the physical parameter space that is consistent with observations. Finally, they will use output from a range of global climate simulations in combination with perturbed physics ensembles to obtain an improved estimate of ice loss from the AIS between now and 2100, with a robust quantification of all model errors and consistent with the observational record. The project will foster international collaboration, bring significant advances in the field with broader societal impact, and establish the PI and his team as world-leaders in the field. Key findings will be published in high impact scientific journals and contribute to future IPCC reports. Through partnerships with NUSTEM and the International Glaciology Society and through project websites, public lectures, school workshops and social media channels, the team will communicate about the wider societal and environmental aspects of sea level rise with key stakeholders in climate policy, young people and their influencers in North East England, and a worldwide audience.