Pith. sign in

REVIEW 22 cited by

ACE: A fast, skillful learned global atmospheric model for climate prediction

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2310.02074 v2 pith:4USEIX6E submitted 2023-10-03 physics.ao-ph cs.LG

classification physics.ao-phcs.LG
keywords climatemodelatmosphericemulatorexistingglobalmoisturenearly
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Existing ML-based atmospheric models are not suitable for climate prediction, which requires long-term stability and physical consistency. We present ACE (AI2 Climate Emulator), a 200M-parameter, autoregressive machine learning emulator of an existing comprehensive 100-km resolution global atmospheric model. The formulation of ACE allows evaluation of physical laws such as the conservation of mass and moisture. The emulator is stable for 100 years, nearly conserves column moisture without explicit constraints and faithfully reproduces the reference model's climate, outperforming a challenging baseline on over 90% of tracked variables. ACE requires nearly 100x less wall clock time and is 100x more energy efficient than the reference model using typically available resources. Without fine-tuning, ACE can stably generalize to a previously unseen historical sea surface temperature dataset.

Discussion (0). Sign in to comment.

Forward citations

Cited by 22 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Spatial Generalization Tests for Machine Learning-based Weather Models to Assess Physical Consistency

    physics.ao-ph 2026-07 conditional novelty 7.0 of 10

    Rotating or reversing the simulated planet reveals that GraphCast and NeuralGCM encode present-day geography rather than spatially invariant physics, while a traditional GCM passes the same tests to numerical precision.

  2. Hard conservation correctors can hide a degrading model when training autoregressive emulators

    physics.ao-ph 2026-07 conditional novelty 7.0 of 10

    Supervising only the corrected output of a hard water-budget corrector hides raw-precipitation amplitude drift, letting the required correction grow from 2% to 24% while delivered fields stay exactly conservative.

  3. Disentangling the effects of sea surface temperature and CO$_2$ in global machine learned weather-climate emulators

    physics.ao-ph 2026-06 unverdicted novelty 7.0 of 10

    Trains ACE emulator on independent SST-CO2 variations plus energy constraint to improve accuracy in decoupled climate forcing scenarios.

  4. Examining Fast Radiatively Driven Responses Using Machine-Learning Weather Emulators

    physics.ao-ph 2026-02 unverdicted novelty 7.0 of 10

    Historically trained ML weather emulators quantify fast precipitation changes from CO2 perturbations and produce results that agree with Earth System Models.

  5. Aircast-Mars: A Mars Foundation Model for Global Weather Forecasting with HEALPix-Aware Convolutions

    physics.ao-ph 2026-06 conditional novelty 6.0 of 10

    A 4.3-million-parameter HEALPix-aware U-Net trained on EMARS reanalysis forecasts Mars temperature and winds stably for 25 hours at ~110 km resolution.

  6. Disentangling the effects of sea surface temperature and CO$_2$ in global machine learned weather-climate emulators

    physics.ao-ph 2026-06 conditional novelty 6.0 of 10

    Training climate emulators on random-CO2 runs that break SST–CO2 correlation, plus an energy constraint, yields a data-efficient model that works on AMIP+4K and abrupt 4xCO2 cases prior ACE models mishandled.

  7. Learning to Emulate Chaos: Adversarial Optimal Transport Regularization

    stat.ML 2026-04 conditional novelty 6.0 of 10

    Adversarial optimal transport objectives jointly learn summary statistics and a chaotic-system emulator from a single noisy trajectory, improving long-term statistical fidelity over handcrafted-feature baselines.

  8. Skillful Global Ocean Emulation and the Role of Correlation-Aware Loss

    physics.ao-ph 2026-04 unverdicted novelty 6.0 of 10

    A GraphCast-based ocean emulator achieves skillful 10-15 day forecasts, with a Mahalanobis loss that accounts for variable correlations improving performance over MSE and acting as a statistical-dynamical regularizer.

  9. A PMP-inspired Evaluation Framework for Assessing Deep-Learning Earth System Models

    physics.ao-ph 2026-04 conditional novelty 6.0 of 10

    Standard PMP diagnostics show DL-ESMs match CMIP-class skill on many large-scale fields but still fail key precipitation, monsoon, and long-run stability tests.

  10. A PMP-inspired Evaluation Framework for Assessing Deep-Learning Earth System Models

    physics.ao-ph 2026-04 unverdicted novelty 6.0 of 10

    The paper presents a PMP-based evaluation framework to test deep-learning Earth system models on climatology and modes of variability using observational data.

  11. Physics constraints and response validation in discrete-time reduced-order modeling: from idealized turbulent systems to climate dynamics

    nlin.CD 2026-02 conditional novelty 6.0 of 10

    A discrete-time energy-conserving neural map with FDT-derived causal regularization reproduces stationary statistics and forced responses of CdV and Lorenz-96 turbulence from unperturbed data alone.

  12. Physics constraints and response validation in discrete-time reduced-order modeling: from idealized turbulent systems to climate dynamics

    nlin.CD 2026-02 unverdicted novelty 6.0 of 10

    A framework builds stable neural models of turbulent dynamics by enforcing energy-preserving nonlinearities and causal constraints in discrete-time flow maps, demonstrated on Charney-DeVore and Lorenz-96 systems.

  13. HealDA: Highlighting the importance of initial errors in end-to-end AI weather forecasts

    physics.ao-ph 2026-01 conditional novelty 6.0 of 10

    HealDA supplies ML-based initial conditions for AI weather models that produce forecasts trailing ERA5-initialized runs by less than one day of effective lead time, with the skill gap arising mainly from initial error size.

  14. MoWE : A Mixture of Weather Experts

    cs.LG 2025-09 conditional novelty 6.0 of 10

    MoWE, a ViT-based gating network, combines forecasts from Pangu, Aurora, and FCN3 with per-grid-point weights and beats each expert and the simple mean in RMSE.

  15. LUCIE-3D: A three-dimensional climate emulator for forced responses

    cs.LG 2025-09 conditional novelty 6.0 of 10

    A lightweight 3D climate emulator trained on 30 years of reanalysis reproduces CO2-driven surface warming and stratospheric cooling with long-term stability.

  16. GEN2: A Generative Prediction-Correction Framework for Long-time Emulations of Spatially-Resolved Climate Extremes

    physics.comp-ph 2025-08 conditional novelty 6.0 of 10

    A generative Gaussian-plus-diffusion emulator trained on one CMIP6 SSP585 realization reproduces extreme temperature statistics under other emission scenarios.

  17. LatticeVision: Image to Image Networks for Modeling Non-Stationary Spatial Data

    stat.ML 2025-05 unverdicted novelty 6.0 of 10

    Image-to-image networks estimate parameters of non-stationary SAR models faster and more accurately than traditional methods by framing fields and parameters as images.

  18. Learning to Emulate Chaos: Adversarial Optimal Transport Regularization

    stat.ML 2026-04 unverdicted novelty 5.0 of 10

    Adversarial optimal transport objectives train neural emulators with improved long-term statistical fidelity on chaotic systems.

  19. The Equilibrium Response of Atmospheric Machine-Learning Models to Uniform Sea Surface Temperature Warming

    physics.ao-ph 2025-10 conditional novelty 5.0 of 10

    Under uniform +2 K SST warming, ML atmospheric emulators reproduce precipitation changes but show deficient land warming, upper-tropospheric warming, and radiative responses compared with the AM4 GCM.

  20. On the Genealogy of Machine Learning Weather Prediction

    physics.ao-ph 2026-07 conditional novelty 4.5 of 10

    ML weather prediction inherited NWP’s IVP framing, so model choice should explicitly align with either physical system structure or data statistical structure rather than defaulting to learned time-steppers.

  21. Assessing Emulator Design and Training for Modal Aerosol Microphysics Parameterizations in E3SMv2

    physics.ao-ph 2026-04 unverdicted novelty 4.0 of 10

    With proper scaling and training convergence, a moderate-sized feedforward neural network can reproduce key aerosol concentration changes from the MAM4 microphysics module in E3SMv2.

  22. A PMP-inspired Evaluation Framework for Assessing Deep-Learning Earth System Models

    physics.ao-ph 2026-04 conditional novelty 4.0 of 10

    A PMP-based evaluation framework for testing deep-learning Earth system models on climate-relevant diagnostics beyond short-range forecasts.

Pith tools