Uncertainty-enabled machine learning for emulation of regional sea-level change caused by the Antarctic Ice Sheet

2024-06-21

arXiv.org

AI Summary

Predicting sea-level response to Antarctic ice-sheet mass loss under climate change is computationally prohibitive using conventional static sea-level equation (SLE) solvers. To address this, we develop 27 region-specific neural network surrogate models for coastal zones, replacing numerical SLE integration. Our method innovatively combines a feedforward neural network with linear regression-based post-processing to enable physically interpretable uncertainty quantification and well-calibrated prediction intervals. Trained on ISMIP6-2100 ice-sheet simulations, the surrogates achieve accuracy comparable to the reference SLE solver while accelerating individual predictions by approximately two orders of magnitude (~100×). The approach thus delivers high computational efficiency, robust reliability, and physical interpretability—establishing a new paradigm for rapid, large-scale assessment of sea-level change impacts.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Computer Vision: Low Level & Physics-based Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines

Abstract

Projecting sea-level change in various climate-change scenarios typically involves running forward simulations of the Earth's gravitational, rotational and deformational (GRD) response to ice mass change, which requires high computational cost and time. Here we build neural-network emulators of sea-level change at 27 coastal locations, due to the GRD effects associated with future Antarctic Ice Sheet mass change over the 21st century. The emulators are based on datasets produced using a numerical solver for the static sea-level equation and published ISMIP6-2100 ice-sheet model simulations referenced in the IPCC AR6 report. We show that the neural-network emulators have an accuracy that is competitive with baseline machine learning emulators. In order to quantify uncertainty, we derive well-calibrated prediction intervals for simulated sea-level change via a linear regression postprocessing technique that uses (nonlinear) machine learning model outputs, a technique that has previously been applied to numerical climate models. We also demonstrate substantial gains in computational efficiency: a feedforward neural-network emulator exhibits on the order of 100 times speedup in comparison to the numerical sea-level equation solver that is used for training.

Problem

• Emulating regional sea-level change from Antarctic ice melt with neural networks
• Quantifying prediction uncertainty in sea-level projections using machine learning
• Accelerating computationally expensive sea-level simulations by 100 times

Innovation

• Neural network emulators replace numerical sea-level simulations
• Linear regression postprocessing quantifies prediction uncertainty
• Feedforward neural networks achieve 100x computational speedup