skip to main content


Title: Statistical Emulation of Ice-Sheet Model Simulations to Estimate Uncertainty in Future Antarctic Sea-Level Contributions
Observational estimates of Antarctic ice loss have accelerated in recent decades, and worst-case scenarios of modeling studies have suggested potentially catastrophic sea level rise (~2 meters) by the end of the century. However, modeled contributions to global mean sea level from the Antarctic ice-sheet (AIS) in the 21st century are highly uncertain, in part because ice-sheet model parameters are poorly constrained. Individual ice-sheet model runs are also deterministic and not computationally efficient enough to generate the continuous probability distributions required for incorporation into a holistic framework of probabilistic sea-level projections. To address these shortfalls, we statistically emulate an ice-sheet model using Gaussian Process (GP) regression. GP modeling is a non-parametric machine-learning technique which maps inputs (e.g. forcing or model parameters) to target outputs (e.g. sea-level contributions from the Antarctic ice-sheet) and has the inherent and important advantage that emulator uncertainty is explicitly quantified. We construct emulators for the last interglacial period and an RCP8.5 scenario, and separately for the western, eastern, and total AIS. Separate emulation of western and eastern AIS is important because their evolutions and physical responses to climate forcing are distinct. The emulators are trained on 196 ensemble members for each scenario, composed by varying the parameters of maximum rate of ice-cliff wastage and the coefficient of hydrofracturing. We condition the emulators on last interglacial proxy sea-level records and modern GRACE measurements and exclude poor-fitting ensemble members. The resulting emulators are sampled to produce probability distributions that fill intermediate gaps between discrete ice-sheet model outcomes. We invert emulated high and low probability sea-level contributions in 2100 to explore 21st century evolution pathways; results highlight the deep uncertainty of ice-sheet model physics and the importance of using observations to narrow the range of parameters. Our approach is designed to be flexible such that other ice-sheet models or parameter spaces may be substituted and explored with the emulator.  more » « less
Award ID(s):
1664013
NSF-PAR ID:
10113504
Author(s) / Creator(s):
Date Published:
Journal Name:
AGU Fall Meeting
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    The Antarctic ice sheet (AIS) will be a dominant contributor to global mean sea level rise in the 21st century but remains a major source of uncertainty. The Ice Sheet Model Intercomparison for CMIP6 (ISMIP6) is an ensemble of continental‐scale models for studying the evolution of the AIS and projecting its future contribution to sea level. Due to their complexity and computational cost, ISMIP6 simulations are sparse and generated infrequently. Emulators are smaller‐scale models that approximate ISMs and enable experimentation and exploration into the drivers of sea level change. We introduce a neural network (NN) emulator to approximate the ISMIP6 ensemble, using a variational Long Short‐Term Memory (LSTM) with Monte Carlo dropout to quantify single‐projection uncertainty. The proposed NN emulator is compared to a Gaussian Process (GP) emulator on four criteria: accuracy of point estimates and predictive distributions of individual model projections, approximation of the ensemble projections, and model training time. The NN predicts more accurately on single projections, with a mean absolute error of 0.46 mm Sea Level Equivalent (SLE) versus 0.73 mm SLE for the GP, and has more accurate uncertainty estimates. The NN emulator also better approximates the ensemble distribution of ISMIP6 model projections, with a Kullback‐Leibler divergence of 18.26 versus 199.14 for GP at the projection year 2100. The NN enables more accurate experimentation with a reduced runtime, offering a new tool for understanding the important role of regional precipitation, ice sheet drainage systems, and interannual and longer timescale dynamics.

     
    more » « less
  2. Previous studies have interpreted Last Interglacial (LIG;129–116 ka) sea‐level estimates in multiple different ways to calibrate projections of future Antarctic ice‐sheet (AIS) mass loss and associated sea‐level rise. This study systematically explores the extent to which LIG constraints could inform future Antarctic contributions to sea‐level rise. We develop a Gaussian process emulator of an ice‐sheet model to produce continuous probabilistic projections of Antarctic sea‐level contributions over the LIG and a future high‐emissions scenario. We use a Bayesian approach conditioning emulator projections on a set of LIG constraints to find associated likelihoods of model parameterizations. LIG estimates inform both the probability of past and future ice‐sheet instabilities and projections of future sea‐level rise through 2150. Although best‐available LIG estimates do not meaningfully constrain Antarctic mass loss projections or physical processes until 2060, they become increasingly informative over the next 130 years. Uncertainties of up to 50 cm remain in future projections even if LIG Antarctic mass loss is precisely known (±5 cm), indicating that there is a limit to how informative the LIG could be for ice‐sheet model future projections. The efficacy of LIG constraints on Antarctic mass loss also depends on assumptions about the Greenland ice sheet and LIG sea‐level chronology. However, improved field measurements and understanding of LIG sea levels still have potential to improve future sea‐level projections, highlighting the importance of continued observational efforts.

     
    more » « less
  3. Abstract. Antarctic ice shelves are vulnerable to warming ocean temperatures, and some have already begun thinning in response to increased basal melt rates.Sea level is therefore expected to rise due to Antarctic contributions, but uncertainties in its amount and timing remain largely unquantified. In particular, there is substantial uncertainty in future basal melt rates arising from multi-model differences in thermal forcing and how melt rates depend on that thermal forcing. To facilitate uncertainty quantification in sea level rise projections, we build, validate, and demonstrate projections from a computationally efficient statistical emulator of a high-resolution (4 km) Antarctic ice sheet model, the Community Ice Sheet Model version 2.1. The emulator is trained to a large (500-member) ensemble of 200-year-long 4 km resolution transient ice sheet simulations, whereby regional basal melt rates are perturbed by idealized (yet physically informed) trajectories. The main advantage of our emulation approach is that by sampling a wide range of possible basal melt trajectories, the emulator can be used to (1) produce probabilistic sea level rise projections over much larger Monte Carlo ensembles than are possible by direct numerical simulation alone, thereby providing better statistical characterization of uncertainties, and (2) predict the simulated ice sheet response under differing assumptions about basal melt characteristics as new oceanographic studies are published, without having to run additional numerical ice sheet simulations. As a proof of concept, we propagate uncertainties about future basal melt rate trajectories, derived from regional ocean models, to generate probabilistic sea level rise estimates for 100 and 200 years into the future. 
    more » « less
  4. Abstract

    Deep uncertainty describes situations when there is either ignorance or disagreement over (1) models used to describe key system processes and (2) probability distributions used to characterize the uncertainty of key variables and parameters. Future projections of Antarctic ice sheet (AIS) mass loss remain characterized by deep uncertainty. This complicates decisions on long‐lived coastal protection projects when determining what margin of safety to implement. If the chosen margin of safety does not properly account for uncertainties in sea level rise, the effectiveness of flood protection could decrease over time, potentially putting lives and properties at a greater risk. To address this issue, we develop a flood damage allowance framework for calculating the height of a flood protection strategy needed to ensure that a given level of financial risk is maintained. The damage allowance framework considers decision maker preferences such as planning horizons, protection strategies, and subjective views of AIS stability. We use Manhattan—with the population and built environment fixed in time—to illustrate how our framework could be used to calculate a range of damage allowances based on multiple plausible scenarios of AIS melt. Under high greenhouse gas emissions, we find that results are sensitive to the selection of the upper limit of AIS contributions to sea level rise. Design metrics that specify financial risk targets, such as expected flood damage, allow for the calculation of avoided flood damages (i.e., benefits) that can be combined with estimates of construction cost and then integrated into existing financial decision‐making approaches (e.g., benefit‐cost analysis).

     
    more » « less
  5. SUMMARY

    Mass loss from polar ice sheets is becoming the dominant contributor to current sea level changes, as well as one of the largest sources of uncertainty in sea level projections. The spatial pattern of sea level change is sensitive to the geometry of ice sheet mass changes, and local sea level changes can deviate from the global mean sea level change due to gravitational, Earth rotational and deformational (GRD) effects. The pattern of GRD sea level change associated with the melting of an ice sheet is often considered to remain relatively constant in time outside the vicinity of the ice sheet. For example, in the sea level projections from the most recent IPCC sixth assessment report (AR6), the geometry of ice sheet mass loss was treated as constant during the 21st century. However, ice sheet simulations predict that the geometry of ice mass changes across a given ice sheet and the relative mass loss from each ice sheet will vary during the coming century, producing patters of global sea level changes that are spatiotemporally variable. We adopt a sea level model that includes GRD effects and shoreline migration to calculate time-varying sea level patterns associated with projections of the Greenland and Antarctic Ice Sheets during the coming century. We find that in some cases, sea level changes can be substantially amplified above the global mean early in the century, with this amplification diminishing by 2100. We explain these differences by calculating the contributions of Earth rotation as well as gravitational and deformational effects to the projected sea level changes separately. We find in one case, for example, that ice gain on the Antarctic Peninsula can cause an amplification of up to 2.9 times the global mean sea level equivalent along South American coastlines due to positive interference of GRD effects. To explore the uncertainty introduced by differences in predicted ice mass geometry, we predict the sea level changes following end-member mass loss scenarios for various regions of the Antarctic Ice Sheet from the ISMIP6 model ensemblely, and find that sea level amplification above the global mean sea level equivalent differ by up to 1.9 times between different ice mass projections along global coastlines outside of Greenland and Antarctica. This work suggests that assessments of future sea level hazard should consider not only the integrated mass changes of ice sheets, but also temporal variations in the geometry of the ice mass changes across the ice sheets. As well, this study highlights the importance of constraining the relative timing of ice mass changes between the Greenland and Antarctic Ice Sheets.

     
    more » « less