arxiv: 2602.16090 · v2 · submitted 2026-02-17 · ⚛️ physics.ao-ph · cs.LG

Recognition: 2 theorem links

· Lean Theorem

Examining Fast Radiatively Driven Responses Using Machine-Learning Weather Emulators

Ankur Mahesh , William D. Collins , Travis A. O'Brien , Paul B. Goddard , Sinclaire Zebaze , Shashank Subramanian , James P.C. Duncan , Oliver Watt-Meyer

show 5 more authors

Boris Bonev Thorsten Kurth Karthik Kashinath Michael S. Pritchard Da Yang

Authors on Pith no claims yet

Pith reviewed 2026-05-15 21:12 UTC · model grok-4.3

classification ⚛️ physics.ao-ph cs.LG

keywords machine learningweather emulatorsfast feedbacksradiative forcingprecipitation responsecarbon dioxideEarth system modelshydrological cycle

0 comments

The pith

Historically trained machine-learning weather emulators reproduce the fast precipitation responses to carbon dioxide changes seen in Earth system models.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that machine-learning emulators trained solely on historical weather data can quantify how precipitation changes in response to new levels of carbon dioxide, capturing fast atmospheric feedbacks that operate on weekly timescales. These feedbacks are already active in the present climate and encoded in the reanalysis data used for training, so the emulators do not require retraining on future conditions. The authors apply the emulators to reduced and elevated CO2 concentrations and find that the resulting precipitation shifts match those produced by full-physics Earth system models. This establishes a route to studying the global hydrological cycle and radiative-convective equilibrium using existing ML tools. The work highlights the distinction between fast feedbacks, which are accessible now, and slower ocean-driven changes that lack historical analogues.

Core claim

We show that the responses from historically trained emulators agree with those produced by full-physics Earth System Models (ESMs). Without retraining on prospective Earth system conditions, we use ML weather emulators to quantify the fast precipitation response to reduced and elevated carbon dioxide concentrations with no recent historical precedent.

What carries the argument

Historically trained ML weather emulators applied directly to altered carbon dioxide concentrations to isolate fast radiative-convective equilibrium responses and precipitation changes.

If this is right

Fast precipitation responses to CO2 can be quantified using existing ML emulators without any retraining on new data.
The global hydrological cycle under radiative perturbations can be examined through these emulators in addition to traditional models.
Agreement with ESMs confirms that historical training data already encodes the relevant fast feedback physics.
ML emulators provide a practical route to studying fast processes in global climate alongside full-physics simulations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same emulators could be tested on fast responses to other short-term forcings such as aerosols or solar changes.
Hybrid setups could combine these fast-response emulators with slower ocean models to span multiple timescales.
This approach may support rapid ensemble exploration of precipitation impacts under varied greenhouse-gas trajectories.
Validation against ESMs for CO2 suggests the emulators might generalize to other novel atmospheric states on weekly timescales.

Load-bearing premise

The physics of fast radiative feedbacks is fully present in historical meteorological reanalyses and remains functional under historical boundary conditions, allowing the emulators to respond correctly to novel CO2 concentrations without retraining.

What would settle it

A side-by-side run in which the emulators' predicted global-mean or regional precipitation change under a specific CO2 perturbation differs substantially in magnitude or pattern from the corresponding ESM output would falsify the reported agreement.

read the original abstract

The response of the climate system to increased greenhouse gases and other radiative perturbations is governed by a combination of fast and slow feedbacks. Slow feedbacks are typically activated in response to changes in ocean temperatures on decadal timescales and manifest as changes in climatic state with no recent historical analogue. However, fast feedbacks are activated in response to rapid atmospheric physical processes on weekly timescales, and they are already operative in the present-day climate. This distinction implies that the physics of fast radiative feedbacks is present in the historical meteorological reanalyses used to train many recent successful machine-learning-based (ML) emulators of weather and climate. In addition, these feedbacks are functional under the historical boundary conditions pertaining to the top-of-atmosphere radiative balance and sea-surface temperatures. Together, these factors imply that we can use historically trained ML weather emulators to study the response of radiative-convective equilibrium (RCE), and hence the global hydrological cycle, to perturbations in carbon dioxide and other well-mixed greenhouse gases. Without retraining on prospective Earth system conditions, we use ML weather emulators to quantify the fast precipitation response to reduced and elevated carbon dioxed concentrations with no recent historical precedent. We show that the responses from historically trained emulators agree with those produced by full-physics Earth System Models (ESMs). In conclusion, we discuss the prospects for and advantages from using ESMs and ML emulators to study fast processes in global climate.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Historically trained ML weather emulators can quantify fast precipitation responses to novel CO2 levels and match ESM outputs.

read the letter

The main thing to know is that historically trained ML weather emulators can be used to study fast precipitation responses to CO2 perturbations without retraining, and the results agree with those from full ESMs. The novelty comes from applying these emulators to radiative perturbations outside the historical range for the hydrological cycle. The authors explain that fast feedbacks are present in reanalysis data and functional under historical conditions, allowing this transfer. They demonstrate it for both reduced and elevated CO2 levels. This is a practical step that could lower the cost of exploring fast climate feedbacks. The work does well by providing the perturbation details and comparison metrics in the full text, which tests the key assumption directly. The agreement with ESMs is shown without circularity, as validation is against independent models. The thinking is straightforward and engages with the distinction between fast and slow processes. Soft spots are limited. The abstract is light on numbers, but the manuscript fills that in. It might be worth checking how robust the results are to different emulator setups or larger perturbations, but nothing load-bearing seems off. This paper is aimed at climate researchers using ML tools for feedback studies. A reader working on atmospheric responses or emulator validation would find it useful. It deserves peer review because the approach is testable and adds a new tool for fast process analysis. I recommend sending it out for review.

Referee Report

0 major / 3 minor

Summary. The paper claims that historically trained machine-learning weather emulators can quantify fast precipitation responses to novel CO2 concentrations (both reduced and elevated) without retraining, because fast radiative feedbacks are already encoded in historical reanalysis data and remain functional under historical boundary conditions. It reports that these emulator-derived responses agree with those produced by full-physics Earth System Models (ESMs) and discusses prospects for hybrid ESM-ML studies of fast processes in radiative-convective equilibrium and the global hydrological cycle.

Significance. If the reported agreement holds under the described perturbation protocol, the work demonstrates a computationally efficient route to isolating fast radiatively driven responses using existing emulators. This could complement traditional ESM experiments by enabling rapid exploration of novel greenhouse-gas forcings while retaining the physics already present in reanalysis-trained models. The approach also supplies a concrete test of whether fast feedbacks can be accessed without retraining on future-like states.

minor comments (3)

[Abstract] Abstract: the statement that emulator responses 'agree with' ESMs would be more informative if it included at least one quantitative metric (e.g., global-mean precipitation change or pattern correlation) together with an indication of uncertainty.
[Methods] The perturbation implementation (exact CO2 concentrations, how they are imposed on the emulator, and any adjustments to sea-surface temperature or top-of-atmosphere balance) should be stated explicitly in the main text with a table or short protocol subsection so that the experiment is fully reproducible from the manuscript alone.
[Figures] Figure captions and axis labels should explicitly note the time scale (e.g., days 1-30) over which the fast response is averaged, to distinguish it clearly from any slow-adjustment component.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive assessment of the work and recommendation for minor revision. No specific major comments were raised in the report, so we have no point-by-point responses to provide at this stage. We will incorporate any minor suggestions from the full review in the revised manuscript.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper trains ML emulators exclusively on historical reanalysis data and then applies them to CO2 perturbations outside the training distribution. The central result is obtained by running these emulators forward and comparing their fast precipitation responses directly to independent full-physics ESM simulations under identical perturbations. This comparison constitutes external validation rather than any internal fit, self-definition, or self-citation chain. No equations, parameter-fitting steps, or uniqueness theorems are shown to reduce the reported agreement to the inputs by construction. The assumption that fast feedbacks are already encoded in historical data is treated as a testable hypothesis, not as a definitional premise. Consequently the derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on two domain assumptions: that fast radiative-convective physics is already encoded in historical reanalysis data used for training, and that emulators can generalize to new top-of-atmosphere radiative balances without retraining.

axioms (2)

domain assumption Fast radiative feedbacks are captured in historical meteorological reanalyses used to train ML emulators.
Explicitly invoked in the abstract as the reason historical training data suffices.
domain assumption ML emulators trained under historical boundary conditions can respond to altered CO2 concentrations.
Required to justify applying the models to reduced and elevated CO2 levels with no recent historical precedent.

pith-pipeline@v0.9.0 · 5610 in / 1195 out tokens · 50430 ms · 2026-05-15T21:12:16.344302+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We use the Allen Institute for Artificial Intelligence (Ai2) Climate Emulator (ACE) [25], an ML emulator that autoregressively predicts the three-dimensional atmospheric state with a 6-hour timestep and 1-degree horizontal resolution... We introduce a modified architecture of ACE that diagnoses precipitation and latent heat flux solely based on the local vertical profiles of the atmospheric state.
IndisputableMonolith/Foundation/AlphaCoordinateFixation.lean alpha_pin_under_high_calibration unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

ACE-RRTMG is a hybrid modeling approach in which ACE predicts the atmospheric time evolution with ML while retaining a physics-based radiative transfer scheme through RRTMG.

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages · 4 internal anchors

[1]

Physics Today42(5), 22–32 (1989) https://doi.org/10.1063/ 1.881167

Ramanathan, V., Barkstrom, B.R., Harrison, E.F.: Climate and the Earth’s Radiation Budget. Physics Today42(5), 22–32 (1989) https://doi.org/10.1063/ 1.881167

work page 1989
[2]

The physics of climate change: simple models in climate science

Jeevanjee, N.: The physics of climate change: simple models in climate science. arXiv (2018). https://doi.org/10.48550/ARXIV.1802.02695

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1802.02695 2018
[3]

Journal of Climate36(11), 3537–3551 (2023) https://doi.org/10.1175/ JCLI-D-22-0513.1

Stjern, C.W., Forster, P.M., Jia, H., Jouan, C., Kasoar, M.R., Myhre, G., Olivie, D., Quaas, J., Samset, B.H., Sand, M., Takemura, T., Voulgarakis, A., Wells, C.D.: The Time Scales of Climate Responses to Carbon Dioxide and Aerosols. Journal of Climate36(11), 3537–3551 (2023) https://doi.org/10.1175/ JCLI-D-22-0513.1

work page 2023
[4]

Geophysical Research Letters43(6), 2782–2791 (2016) https: //doi.org/10.1002/2016GL068064

Samset, B.H., Myhre, G., Forster, P.M., Hodnebrog, O., Andrews, T., Faluvegi, G., Flaeschner, D., Kasoar, M., Kharin, V., Kirkevag, A., Lamarque, J.-F., Olivie, D., Richardson, T., Shindell, D., Shine, K.P., Takemura, T., Voulgarakis, A.: Fast and slow precipitation responses to individual climate forcers: A PDRMIP multimodel study. Geophysical Research L...

work page doi:10.1002/2016gl068064 2016
[5]

Geophysical Research Letters 35(4) (2008) https://doi.org/10.1029/2007gl032273

Andrews, T., Forster, P.M.: CO 2 forcing induces semi-direct effects with consequences for climate feedback interpretations. Geophysical Research Letters 35(4) (2008) https://doi.org/10.1029/2007gl032273

work page doi:10.1029/2007gl032273 2008
[6]

Journal of Climate21(1), 58–71 (2008) https://doi.org/10.1175/ 2007jcli1834.1

Gregory, J., Webb, M.: Tropospheric Adjustment Induces a Cloud Component in CO2 Forcing. Journal of Climate21(1), 58–71 (2008) https://doi.org/10.1175/ 2007jcli1834.1

work page 2008
[7]

Journal of Climate22(10), 2557–2570 (2009) https://doi.org/ 10.1175/2008jcli2759.1

Andrews, T., Forster, P.M., Gregory, J.M.: A Surface Energy Perspective on Climate Change. Journal of Climate22(10), 2557–2570 (2009) https://doi.org/ 10.1175/2008jcli2759.1

work page doi:10.1175/2008jcli2759.1 2009
[8]

Geophysical Research Letters 37(14) (2010) https://doi.org/10.1029/2010gl043991

Andrews, T., Forster, P.M., Boucher, O., Bellouin, N., Jones, A.: Precipitation, radiative forcing and global temperature change. Geophysical Research Letters 37(14) (2010) https://doi.org/10.1029/2010gl043991

work page doi:10.1029/2010gl043991 2010
[9]

Current Climate Change Reports1(2), 103–113 (2015) https://doi.org/10.1007/ s40641-015-0007-5

Kamae, Y., Watanabe, M., Ogura, T., Yoshimori, M., Shiogama, H.: Rapid Adjustments of Cloud and Hydrological Cycle to Increasing CO 2: a Review. Current Climate Change Reports1(2), 103–113 (2015) https://doi.org/10.1007/ s40641-015-0007-5

work page 2015
[10]

International Journal of Climatology37(3), 1156–1166 (2017) https://doi.org/10.1002/joc.4763 40

Tian, D., Dong, W., Gong, D., Guo, Y., Yang, S.: Fast responses of climate system to carbon dioxide, aerosols and sulfate aerosols without the mediation of SST in the CMIP5. International Journal of Climatology37(3), 1156–1166 (2017) https://doi.org/10.1002/joc.4763 40

work page doi:10.1002/joc.4763 2017
[11]

Climate Dynamics56(3–4), 1083–1104 (2020) https://doi.org/10.1007/s00382-020-05522-w

Douville, H., John, A.: Fast adjustment versus slow SST-mediated response of daily precipitation statistics to abrupt 4xCO2. Climate Dynamics56(3–4), 1083–1104 (2020) https://doi.org/10.1007/s00382-020-05522-w

work page doi:10.1007/s00382-020-05522-w 2020
[12]

Journal of Climate29(2), 583–594 (2016) https://doi.org/10.1175/jcli-d-15-0174

Richardson, T.B., Forster, P.M., Andrews, T., Parker, D.J.: Understanding the Rapid Precipitation Response to CO2 and Aerosol Forcing on a Regional Scale*. Journal of Climate29(2), 583–594 (2016) https://doi.org/10.1175/jcli-d-15-0174. 1

work page doi:10.1175/jcli-d-15-0174 2016
[13]

Part I: Transient Adjustment

Dong, B., Gregory, J.M., Sutton, R.T.: Understanding Land–Sea Warming Contrast in Response to Increasing Greenhouse Gases. Part I: Transient Adjustment. Journal of Climate22(11), 3079–3097 (2009) https://doi.org/10. 1175/2009jcli2652.1

work page 2009
[14]

Geoscientific Model Development9(5), 1937–1958 (2016) https://doi.org/10.5194/gmd-9-1937-2016

Eyring, V., Bony, S., Meehl, G.A., Senior, C.A., Stevens, B., Stouffer, R.J., Taylor, K.E.: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geoscientific Model Development9(5), 1937–1958 (2016) https://doi.org/10.5194/gmd-9-1937-2016

work page doi:10.5194/gmd-9-1937-2016 1937
[15]

Geoscientific Model Development18(17), 5575–5603 (2025) https://doi.org/10

Mahesh, A., Collins, W.D., Bonev, B., Brenowitz, N., Cohen, Y., Elms, J., Harrington, P., Kashinath, K., Kurth, T., North, J., O’Brien, T., Pritchard, M., Pruitt, D., Risser, M., Subramanian, S., Willard, J.: Huge ensembles – Part 1: Design of ensemble weather forecasts using spherical Fourier neural operators. Geoscientific Model Development18(17), 5575–...

work page 2025
[16]

Nature619(7970), 533–538 (2023) https://doi.org/10.1038/s41586-023-06185-3

Bi, K., Xie, L., Zhang, H., Chen, X., Gu, X., Tian, Q.: Accurate medium-range global weather forecasting with 3D neural networks. Nature619(7970), 533–538 (2023) https://doi.org/10.1038/s41586-023-06185-3

work page doi:10.1038/s41586-023-06185-3 2023
[17]

Science382(6677), 1416–1421 (2023) https://doi.org/10.1126/science.adi2336

Lam, R., Sanchez-Gonzalez, A., Willson, M., Wirnsberger, P., Fortunato, M., Alet, F., Ravuri, S., Ewalds, T., Eaton-Rosen, Z., Hu, W., Merose, A., Hoyer, S., Holland, G., Vinyals, O., Stott, J., Pritzel, A., Mohamed, S., Battaglia, P.: Learning skillful medium-range global weather forecasting. Science382(6677), 1416–1421 (2023) https://doi.org/10.1126/sci...

work page doi:10.1126/science.adi2336 2023
[18]

Artificial Intelligence for the Earth Systems4(2) (2025) https://doi.org/10.1175/aies-d-24-0061.1

Willard, J.D., Harrington, P., Subramanian, S., Mahesh, A., O’Brien, T.A., Collins, W.D.: Analyzing and Exploring Training Recipes for Large-Scale Transformer-Based Weather Prediction. Artificial Intelligence for the Earth Systems4(2) (2025) https://doi.org/10.1175/aies-d-24-0061.1

work page doi:10.1175/aies-d-24-0061.1 2025
[19]

arXiv (2023)

Bonev, B., Kurth, T., Hundt, C., Pathak, J., Baust, M., Kashinath, K., Anandkumar, A.: Spherical Fourier Neural Operators: Learning Stable Dynamics on the Sphere. arXiv (2023). https://doi.org/10.48550/ARXIV.2306.03838

work page doi:10.48550/arxiv.2306.03838 2023
[20]

FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators

Pathak, J., Subramanian, S., Harrington, P., Raja, S., Chattopadhyay, A., Mardani, M., Kurth, T., Hall, D., Li, Z., Azizzadenesheli, K., Hassanzadeh, 41 P., Kashinath, K., Anandkumar, A.: FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators. arXiv (2022). https://doi.org/10.48550/ARXIV.2202.11214

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2202.11214 2022
[21]

Collins, Michael S

Bonev, B., Kurth, T., Mahesh, A., Bisson, M., Kossaifi, J., Kashinath, K., Anandkumar, A., Collins, W.D., Pritchard, M.S., Keller, A.: FourCastNet 3: A geometric approach to probabilistic machine-learning weather forecasting at scale. arXiv (2025). https://doi.org/10.48550/ARXIV.2507.12144

work page doi:10.48550/arxiv.2507.12144 2025
[22]

Watt-Meyer, G

Watt-Meyer, O., Dresdner, G., McGibbon, J., Clark, S.K., Henn, B., Duncan, J., Brenowitz, N.D., Kashinath, K., Pritchard, M.S., Bonev, B., Peters, M.E., Bretherton, C.S.: ACE: A fast, skillful learned global atmospheric model for climate prediction (2023). https://arxiv.org/abs/2310.02074

work page arXiv 2023
[23]

AGU Advances6(4) (2025) https://doi.org/10.1029/ 2025av001706

Cresswell-Clay, N., Liu, B., Durran, D.R., Liu, Z., Espinosa, Z.I., Moreno, R.A., Karlbauer, M.: A Deep Learning Earth System Model for Efficient Simulation of the Observed Climate. AGU Advances6(4) (2025) https://doi.org/10.1029/ 2025av001706

work page 2025
[24]

Chapman, J

Chapman, W.E., Schreck, J.S., Sha, Y., Gagne, D.J., Kimpara, D., Zanna, L., Mayer, K.J., Berner, J.: CAMulator: Fast Emulation of the Community Atmosphere Model. arXiv (2025). https://doi.org/10.48550/ARXIV.2504.06007

work page doi:10.48550/arxiv.2504.06007 2025
[25]

Journal of Geophysical Research: Machine Learning and Computation1(3) (2024) https: //doi.org/10.1029/2024jh000136

Duncan, J.P.C., Wu, E., Golaz, J., Caldwell, P.M., Watt-Meyer, O., Clark, S.K., McGibbon, J., Dresdner, G., Kashinath, K., Bonev, B., Pritchard, M.S., Bretherton, C.S.: Application of the AI2 Climate Emulator to E3SMv2’s Global Atmosphere Model, With a Focus on Precipitation Fidelity. Journal of Geophysical Research: Machine Learning and Computation1(3) (...

work page doi:10.1029/2024jh000136 2024
[26]

Journal of Advances in Modeling Earth Systems14(12) (2022) https://doi.org/10.1029/2022ms003156

Golaz, J., Van Roekel, L.P., Zheng, X., Roberts, A.F., Wolfe, J.D., Lin, W., Bradley, A.M., Tang, Q., Maltrud, M.E., Forsyth, R.M., Zhang, C., Zhou, T., Zhang, K., Zender, C.S., Wu, M., Wang, H., Turner, A.K., Singh, B., Richter, J.H., Qin, Y., Petersen, M.R., Mametjanov, A., Ma, P., Larson, V.E., Krishna, J., Keen, N.D., Jeffery, N., Hunke, E.C., Hannah,...

work page doi:10.1029/2022ms003156 2022
[27]

Journal of Geophysical Research: Atmospheres113(D13) (2008) https://doi.org/10.1029/2008jd009944

Iacono, M.J., Delamere, J.S., Mlawer, E.J., Shephard, M.W., Clough, S.A., Collins, W.D.: Radiative forcing by long-lived greenhouse gases: Calculations 42 with the AER radiative transfer models. Journal of Geophysical Research: Atmospheres113(D13) (2008) https://doi.org/10.1029/2008jd009944

work page doi:10.1029/2008jd009944 2008
[28]

Journal of Climate35(13), 4027–4047 (2022) https://doi.org/10.1175/jcli-d-21-0275.1

Romps, D.M., Seeley, J.T., Edman, J.P.: Why the Forcing from Carbon Dioxide Scales as the Logarithm of Its Concentration. Journal of Climate35(13), 4027–4047 (2022) https://doi.org/10.1175/jcli-d-21-0275.1

work page doi:10.1175/jcli-d-21-0275.1 2022
[29]

Balaji, V., Couvreux, F., Deshayes, J., Gautrais, J., Hourdin, F., Rio, C.: Are general circulation models obsolete? Proceedings of the National Academy of Sciences119(47) (2022) https://doi.org/10.1073/pnas.2202075119

work page doi:10.1073/pnas.2202075119 2022
[30]

https://onlinelibrary.wiley.com/doi/abs/10.1111/j.2153-3490.1969.tb00466.x (accessed 6 April 2025)

Lorenz, E.N.: The predictability of a flow which possesses many scales of motion. Tellus21(3), 289–307 (1969) https://doi.org/10.1111/j.2153-3490.1969.tb00444. x

work page doi:10.1111/j.2153-3490.1969.tb00444 1969
[31]

Charney, J.G.: The feasibility of a global observation and analysis experiment. Bull. Amer. Meteor. Soc.47, 200–230 (1966)

work page 1966
[32]

Nature632(8027), 1060–1066 (2024) https://doi.org/10

Kochkov, D., Yuval, J., Langmore, I., Norgaard, P., Smith, J., Mooers, G., Kl¨ ower, M., Lottes, J., Rasp, S., D¨ uben, P., Hatfield, S., Battaglia, P., Sanchez-Gonzalez, A., Willson, M., Brenner, M.P., Hoyer, S.: Neural general circulation models for weather and climate. Nature632(8027), 1060–1066 (2024) https://doi.org/10. 1038/s41586-024-07744-y

work page 2024
[33]

arXiv (2022)

Keisler, R.: Forecasting Global Weather with Graph Neural Networks. arXiv (2022). https://doi.org/10.48550/ARXIV.2202.07575

work page doi:10.48550/arxiv.2202.07575 2022
[34]

Brenowitz, Yair Cohen, Jaideep Pathak, Ankur Mahesh, Boris Bonev, Thorsten Kurth, Dale R

Brenowitz, N.D., Cohen, Y., Pathak, J., Mahesh, A., Bonev, B., Kurth, T., Durran, D.R., Harrington, P., Pritchard, M.S.: A Practical Probabilistic Benchmark for AI Weather Models. Geophysical Research Letters52(7) (2025) https://doi.org/10.1029/2024gl113656

work page doi:10.1029/2024gl113656 2025
[35]

arXiv (2024)

Lang, S., Alexe, M., Clare, M.C.A., Roberts, C., Adewoyin, R., Bouall` egue, Z.B., Chantry, M., Dramsch, J., Dueben, P.D., Hahner, S., Maciel, P., Prieto-Nemesio, A., O’Brien, C., Pinault, F., Polster, J., Raoult, B., Tietsche, S., Leutbecher, M.: AIFS-CRPS: Ensemble forecasting using a model trained with a loss function based on the Continuous Ranked Pro...

work page doi:10.48550/arxiv.2412.15832 2024
[36]

arXiv (2025)

Perkins, W.A., Kwa, A., McGibbon, J., Arcomano, T., Clark, S.K., Watt-Meyer, O., Bretherton, C.S., Harris, L.M.: HiRO-ACE: Fast and skillful AI emulation and downscaling trained on a 3 km global storm-resolving model. arXiv (2025). https://doi.org/10.48550/ARXIV.2512.18224

work page doi:10.48550/arxiv.2512.18224 2025
[37]

5194/egusphere-2024-3874

Dunne, J.P., Hewitt, H.T., Arblaster, J., Bonou, F., Boucher, O., Cavazos, T., Durack, P.J., Hassler, B., Juckes, M., Miyakawa, T., Mizielinski, M., Naik, V., Nicholls, Z., O’Rourke, E., Pincus, R., Sanderson, B.M., Simpson, I.R., Taylor, 43 K.E.: An evolving Coupled Model Intercomparison Project phase 7 (CMIP7) and Fast Track in support of future climate...

work page 2024
[38]

Journal of Advances in Modeling Earth Systems16(8) (2024) https://doi.org/10.1029/2023ms004021

Karlbauer, M., Cresswell-Clay, N., Durran, D.R., Moreno, R.A., Kurth, T., Bonev, B., Brenowitz, N., Butz, M.V.: Advancing Parsimonious Deep Learning Weather Prediction Using the HEALPix Mesh. Journal of Advances in Modeling Earth Systems16(8) (2024) https://doi.org/10.1029/2023ms004021

work page doi:10.1029/2023ms004021 2024
[39]

Learning to Advect: A Neural Semi-Lagrangian Architecture for Weather Forecasting

Pereira, C.A., Gaudreault, S., Dallerit, V., Subich, C., Panday, S., Wei, S., Zhang, S., Rout, S., Haber, E., Spiteri, R.J., Millard, D., Diaconescu, E.: Learning to Advect: A Neural Semi-Lagrangian Architecture for Weather Forecasting. arXiv (2026). https://doi.org/10.48550/ARXIV.2601.21151 . https://arxiv.org/ abs/2601.21151

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2601.21151 2026
[40]

npj Climate and Atmospheric Science8(1), 205 (2025) https://doi.org/10.1038/ s41612-025-01090-0

Watt-Meyer, O., Henn, B., McGibbon, J., Clark, S.K., Kwa, A., Perkins, W.A., Wu, E., Harris, L., Bretherton, C.S.: ACE2: accurately learning subseasonal to decadal atmospheric variability and forced responses. npj Climate and Atmospheric Science8(1), 205 (2025) https://doi.org/10.1038/ s41612-025-01090-0

work page 2025
[41]

Science Advances12(2) (2026) https://doi.org/10

Yuval, J., Langmore, I., Kochkov, D., Hoyer, S.: Neural general circulation models for modeling precipitation. Science Advances12(2) (2026) https://doi.org/10. 1126/sciadv.adv6891

work page 2026
[42]

Journal of Geophysical Research: Machine Learning and Computation2(4) (2025) https://doi.org/10.1029/2024jh000575

Clark, S.K., Watt-Meyer, O., Kwa, A., McGibbon, J., Henn, B., Perkins, W.A., Wu, E., Harris, L.M., Bretherton, C.S.: ACE2-SOM: Coupling an ML Atmospheric Emulator to a Slab Ocean and Learning the Sensitivity of Climate to Changed CO2. Journal of Geophysical Research: Machine Learning and Computation2(4) (2025) https://doi.org/10.1029/2024jh000575

work page doi:10.1029/2024jh000575 2025
[43]

org/10.5194/egusphere-2025-6266

Davenport, E.H., Madan, J.V., Gjini, R., Brzenski, J., Ho, N., Hsu, T.-Y., Liang, Y., Liu, Z., Manivannan, V., Pham, E., Vutukuru, R., Williams, A.I.L., Yang, Z., Yu, R., Lutsko, N.J., Hoyer, S., Watson-Parris, D.: JCM v1.0: A Differentiable, Intermediate-Complexity Atmospheric Model (2026) https://doi. org/10.5194/egusphere-2025-6266

work page doi:10.5194/egusphere-2025-6266 2026
[44]

Mahesh, A., D. Collins, W., Bonev, B., Brenowitz, N., Cohen, Y., Harrington, P., Kashinath, K., Kurth, T., North, J., O’Brien, T.A., Pritchard, M., Pruitt, D., Risser, M., Subramanian, S., Willard, J.: Huge ensembles – Part 2: Properties of a huge ensemble of hindcasts generated with spherical Fourier neural operators. Geoscientific Model Development18(17...

work page 2025
[45]

SamudrACE: Fast and accurate coupled climate modeling with 3D ocean and atmosphere emulators

Duncan, J.P.C., Wu, E., Dheeshjith, S., Subel, A., Arcomano, T., Clark, S.K., Henn, B., Kwa, A., McGibbon, J., Perkins, W.A., Gregory, W., Fernandez-Granda, C., Busecke, J., Watt-Meyer, O., Hurlin, W.J., Adcroft, A., 44 Zanna, L., Bretherton, C.: SamudrACE: Fast and Accurate Coupled Climate Modeling with 3D Ocean and Atmosphere Emulators. arXiv (2025). ht...

work page doi:10.48550/arxiv.2509.12490 2025
[46]

Geophysical Research Letters52(10) (2025) https://doi.org/10.1029/2024gl114318

Dheeshjith, S., Subel, A., Adcroft, A., Busecke, J., Fernandez-Granda, C., Gupta, S., Zanna, L.: Samudra: An AI Global Ocean Emulator for Climate. Geophysical Research Letters52(10) (2025) https://doi.org/10.1029/2024gl114318

work page doi:10.1029/2024gl114318 2025
[47]

Journal of Quantitative Spectroscopy and Radiative Transfer91(2), 233–244 (2005) https://doi.org/10.1016/j.jqsrt

Clough, S.A., Shephard, M.W., Mlawer, E.J., Delamere, J.S., Iacono, M.J., Cady-Pereira, K., Boukabara, S., Brown, P.D.: Atmospheric radiative transfer modeling: a summary of the AER codes. Journal of Quantitative Spectroscopy and Radiative Transfer91(2), 233–244 (2005) https://doi.org/10.1016/j.jqsrt. 2004.05.058

work page doi:10.1016/j.jqsrt 2005
[48]

0.4.0) and climt (v

Monteiro, J.M., McGibbon, J., Caballero, R.: sympl (v. 0.4.0) and climt (v. 0.15.3) – towards a flexible framework for building model hierarchies in Python. Geoscientific Model Development11(9), 3781–3794 (2018) https://doi.org/10. 5194/gmd-11-3781-2018

work page 2018
[49]

Journal of Advances in Modeling Earth Systems11(10), 3074–3089 (2019) https://doi.org/10.1029/ 2019ms001621

Pincus, R., Mlawer, E.J., Delamere, J.S.: Balancing Accuracy, Efficiency, and Flexibility in Radiation Calculations for Dynamical Models. Journal of Advances in Modeling Earth Systems11(10), 3074–3089 (2019) https://doi.org/10.1029/ 2019ms001621

work page 2019
[50]

In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A

Ramaswamy, V., Boucher, O., Haigh, J., Hauglustaine, D., Haywood, J.M., Myhre, G., Nakajima, T., Shi, G.Y., Solomon, S.: Radiative Forcing of Climate Change. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A. (eds.) Climate Change 2001: The Scientific Basis. Contribution of Working Group I to ...

work page 2001
[51]

Journal of the Atmospheric Sciences78(1), 269–286 (2021) https://doi.org/10.1175/jas-d-20-0015.1

Bloxam, K., Huang, Y.: Radiative Relaxation Time Scales Quantified from Sudden Stratospheric Warmings. Journal of the Atmospheric Sciences78(1), 269–286 (2021) https://doi.org/10.1175/jas-d-20-0015.1

work page doi:10.1175/jas-d-20-0015.1 2021
[52]

arXiv (2022)

Ocampo, J., Price, M.A., McEwen, J.D.: Scalable and Equivariant Spherical CNNs by Discrete-Continuous (DISCO) Convolutions. arXiv (2022). https://doi. org/10.48550/ARXIV.2209.13603

work page doi:10.48550/arxiv.2209.13603 2022
[53]

Artificial Intelligence for the Earth Systems3(3) (2024) https://doi

You, H., Wang, J., Wong, R.K.W., Schumacher, C., Saravanan, R., Jun, M.: Prediction of Tropical Pacific Rain Rates with Overparameterized Neural Networks. Artificial Intelligence for the Earth Systems3(3) (2024) https://doi. org/10.1175/aies-d-23-0083.1

work page doi:10.1175/aies-d-23-0083.1 2024
[54]

In: 105th Annual AMS Meeting 2025, vol

Chen, C.-J., Saravanan, R., Szunyogh, I., Schumacher, C.J., Wong, R.: 45 Data-driven models for precipitation forecasting: A two-step approach. In: 105th Annual AMS Meeting 2025, vol. 105, p. 453577 (2025)

work page 2025
[55]

& Mishra, M

Mitra, P.P., Ramavajjala, V.: Learning to forecast diagnostic parameters using pre-trained weather embedding. arXiv (2023). https://doi.org/10.48550/ARXIV. 2312.00290 46

work page internal anchor Pith review doi:10.48550/arxiv 2023