Pith. sign in

REVIEW 3 major objections 5 minor 41 references

Stochastic Emulation of a Fully Coupled Preindustrial E3SMv3 Simulation

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A stochastic coupled emulator reproduces a reference climate model's mean state and centuries-long internal variability, with biases smaller than the model's own errors against observations.

desk verdict A strong, honest coupled-emulator study whose central claim holds up, but whose abstract overstates the variability result given that only the atmosphere carries stochasticity. read the letter →

arxiv 2608.10277 v1 pith:RSXMXEEP submitted 2026-08-10 physics.ao-ph cs.LG

Elynn Wu , James P. C. Duncan , Troy Arcomano , Jeremy McGibbon , Oliver Watt-Meyer , Christopher S. Bretherton , Naser Mahfouz , Claudia Tebaldi
show 6 more authors
Luke Van Roekel Andrew Roberts Wuyin Lin Finn Rebassoo Jean-Christophe Golaz Peter M. Caldwell
This is my paper · ORCID
classification physics.ao-phcs.LG
keywords stochasticclimateemulationcoupledatmosphere-oceanemulatorE3SMv3internalvariabilityENSOprecipitationextremesseaicemachinelearningmodels
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper sets out to show that a fully coupled AI emulator of a climate model can reproduce not only the time-mean climate but also the internal variability that matters for long-term behavior, if the atmosphere component is stochastic rather than deterministic. The authors train the coupled system on 105 years of a pre-industrial control run of a reference earth system model and test it on 400 independent years. On the mean state, the emulator's root-mean-square biases against the reference model (0.62 K surface temperature, 0.19 mm/day precipitation) are much smaller than the reference model's own biases against observations (1.13 K, 1.04 mm/day). With stochastic training the emulator keeps ENSO spectral power, Gulf Stream SST variance, and marginal-ice-zone sea ice variability closer to the reference, while deterministic training tends to damp or collapse that variability. If correct, this makes cheap, fast stochastic emulators a credible tool for generating the long ensembles needed to sample internal climate variability.

What carries the argument

The load-bearing mechanism is ACE2S, a stochastic version of an atmospheric emulator that outputs an ensemble of equally likely states rather than one deterministic forecast, trained with a probabilistic loss combining the fair continuous ranked probability score and a spectral energy score. During coupled fine-tuning, its 5-day mean surface fluxes and wind stress drive two parallel Samudra ocean rollouts, and the ocean returns SST and sea ice fraction. No noise is injected inside Samudra; all stochasticity enters through the atmosphere. This mechanism carries the argument because it is what distinguishes the stochastic runs from the deterministic baseline and what recovers ENSO and eddy variability.

What would settle it

Add stochastic noise inside the Samudra ocean emulator and retrain under the same protocol; if the Gulf Stream SST anomaly standard deviation rises from 0.74 K toward the target 1.27 K and the marginal-ice-zone sea ice variance ratios rise from 0.70 and 0.76 toward 1, the atmosphere-only assumption is falsified. Conversely, finding a deterministic training seed that yields a realistic Niño 3.4 spectrum would weaken the claim that stochasticity is necessary.

Watch

Extended reading notes

Core claim

The central discovery is that replacing a deterministic atmospheric emulator with a stochastic one, and fine-tuning the coupled atmosphere–ocean system with a probabilistic scoring rule, turns the atmosphere into a source of internal variability for the ocean. The resulting system, SamudrACE-E3SMv3 (the Samudra full-depth ocean emulator coupled to the ACE2S stochastic atmosphere emulator), produces a 400-year free-running simulation with no drift in its precipitation distribution, reproduces the Niño 3.4 power spectrum within the spread of the reference model's own 40-year blocks, and recovers about half the Gulf Stream SST variance that the deterministic baseline loses (0.74 K versus 1.27 K anomaly standard deviation in the target). Biases against the reference are smaller than the reference's biases against observations. The paper also documents what is not captured: the emulator underestimates the rarest tropical daily precipitation extremes above roughly 150 mm/day.

Load-bearing premise

The load-bearing premise is that atmosphere-driven stochasticity alone is sufficient to generate the ocean's internal variability; if ocean-internal noise also matters, the remaining variance gaps may not close.

Editorial extensions

If this is right

  • A 400-year free-running stochastic rollout shows no precipitation drift, so the emulator can stand in for the reference control climate for centennial variability studies.
  • Because emulator-to-reference mean biases (0.62 K, 0.19 mm/day) are smaller than reference-to-observation biases (1.13 K, 1.04 mm/day), the emulator is a valid substitute for the model in model-vs-observation comparisons.
  • Stochastic training gives ENSO block-to-block spectral spread comparable to the reference, meaning the emulator reproduces not just the mean spectral peak but the intrinsic randomness of ENSO.
  • The deterministic baseline's variability collapse is invisible in mean-state metrics, so variability-aware diagnostics are needed to select among emulator training seeds.
  • Daily precipitation is reliable to the 99.99th percentile globally and into the far tail over CONUS; only tropical extremes above about 150 mm/day are under-represented.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An implication not drawn by the paper: the failure mode of one deterministic seed, a collapsed overly regular oscillation with normal mean-state skill, suggests deterministic MSE training has multiple solutions with equal mean fidelity, so variability diagnostics should be part of checkpoint and seed selection.
  • A natural testable extension: injecting noise inside the ocean emulator and comparing the Gulf Stream and sea-ice variance gaps would isolate how much missing variance is due to absent ocean-internal stochasticity rather than to atmosphere forcing.
  • The tropical precipitation tail deficit, present in both training and evaluation, points to a concrete training fix, such as tail-weighted or up-sampled losses, that could be tested on the same 400-year evaluation protocol.
  • If the stochastic coupling recipe transfers to another climate model without re-architecting, it may become a general method for producing cheap internal-variability ensembles; the transfer to a second GCM here is a first step toward that conclusion.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. SamudrACE-E3SMv3 couples the stochastic ACE2S atmosphere emulator with the Samudra ocean/sea-ice emulator, both pretrained on 105 years of E3SMv3 preindustrial control output and then fine-tuned jointly with a probabilistic (CRPS + spectral energy score) objective. The coupled emulator is evaluated on a 400-year independent segment of the same E3SMv3 control run. The paper reports climatological RMSBs of 0.62 K in surface temperature and 0.19 mm/day in precipitation, well below E3SMv3's own biases against observations, and shows that the emulator reproduces the daily precipitation distribution out to the 99.99th percentile except for rare tropical extremes. Relative to a deterministic baseline, the stochastic emulator better sustains ENSO spectral power, Gulf Stream SST variance, and marginal-ice-zone sea-ice variability. Code, trained weights, and data are publicly archived.

Significance. If the main claims hold, this is an important step toward fast, fully coupled climate emulation: a roughly 40x speedup on a single GPU with century-scale stability would enable large ensembles and rapid model iteration. The paper's evaluation design is a strength: a held-out 400-year stationary segment, block-averaged spectral uncertainties, and comparisons against E3SMv3's own 40-year block spread. The authors are also commendably honest about limitations (Nordic Seas bias, tropical precipitation tail, residual SST and sea-ice variance deficits, and the open question of ocean-internal noise). The open-data and open-code practices are exemplary. However, the central claim that stochastic training—rather than the concurrent change to a probabilistic loss or the particular random seed—is responsible for the variability improvements is not fully established by the two-seed comparison, and the remaining variability gaps are in fields where ocean-internal noise would be expected to matter. These issues are correctable with additional experiments or careful hedging, so the paper warrants revision rather than rejection.

major comments (3)
  1. [§3.3, Fig S2] The claim that stochastic training is more reliable than deterministic training for ENSO variability rests on only two random seeds per configuration. One deterministic seed produces a plausible spectrum while the other collapses onto an overly regular oscillation, and both stochastic seeds are plausible. With n=2, the collapsed deterministic run could be an unlucky seed rather than evidence of a systematic property of deterministic training, and §3.3's statement that 'We attribute the more realistic ENSO variability ... to the use of a stochastic rather than deterministic atmospheric emulator' goes beyond what these data establish. I recommend either training additional seeds (even three per condition) or reframing the claim as a case study and explicitly stating that the seed count is small. This is load-bearing because the headline 'stochastic training maintains internal variability' is the paper's central novelty and the only direct evidence for it is this comparison.
  2. [§2.2.4, §3.4] The stochastic-versus-deterministic comparison changes more than the presence of stochasticity: the stochastic configuration uses ACE2S with a CRPS + energy-score loss and randomly sampled loss windows, while the deterministic baseline uses ACE2 with an MSE loss and fixed four-ocean-step / two-atmosphere-step windows. The attribution in §3.4 that 'Adding stochasticity to the emulator recovers much of this missing variance' therefore conflates stochasticity with the probabilistic objective and the rollout-sampling scheme. A cleaner attribution would require a deterministic run trained with the same probabilistic loss (or a stochastic run trained with MSE), or at least an explicit caveat that the improvement may reflect the combination of these changes. This matters because §4's mechanistic statement that 'the deterministic ocean inheriting its variability from the stochastic atmosphere' presupposes that the improvement is due to the stochastic source rather than the loss function.
  3. [§2.2.4, §3.4, §4, Abstract] The paper's own reported deficits are concentrated where internally generated ocean and sea-ice variability should be important: Gulf Stream per-gridpoint SST anomaly standard deviation is 0.74 K versus 1.27 K in E3SMv3, and marginal-ice-zone sea-ice anomaly ratios reach only 0.70 (NH) and 0.76 (SH) of target. Because no noise is injected into Samudra and all stochasticity enters through ACE2S's 5-day mean surface forcing, a substantial fraction of the missing variability may be irreducible with the current architecture if E3SMv3's ocean/sea-ice internal variability is partly generated within those components. The authors acknowledge this in §4 ('the importance of such ocean-internal noise ... remains be explored'), but the Abstract's closing claim that 'stochastic coupled emulators can reproduce long-timescale variability with high fidelity' overstates the evidence. I recommend softening that sentence to describe improvement relative to the deterministic baseline rather than high-fidelity reproduction, and explicitly noting that the demonstrated variability is atmosphere-forced only.
minor comments (5)
  1. [§4] There is a typo in the Conclusions: 'variaiblity' should be 'variability', and 'remains be explored' should be 'remains to be explored'.
  2. [§3.1] The sentence 'further displace the ice edge' should read 'further displaces the ice edge'.
  3. [§2.2.2] The total energy budget correction imposes a fitted constant residual of 0.09 W/m2; consider clarifying that the model enforces energy balance only up to this imposed residual, not exact conservation.
  4. [Fig 2 caption] The observation datasets cover different periods (1870-1900 for HadISST and 1985-2014 for GPCP); the text states this, but the caption should also restate the periods or note that they are defined in the text.
  5. [Text S2] The spectral land-fill diagnostic parameters (n=4, k=5, sigma=1) are presented as defaults; a brief sensitivity check would help show that the reported spectral metrics are robust to these choices.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the emulator is trained on one segment of the E3SMv3 control simulation and evaluated on a held-out 400-year segment, which is a standard fidelity benchmark rather than a derivation from the target result.

full rationale

The paper's central claim is that SamudrACE-E3SMv3 reproduces E3SMv3's mean state, variability, and precipitation distribution. This is an emulation benchmark, not a derivation: the model is trained on 105 years (90 train / 5 validation / 10 test) of a re-run of the E3SMv3 pre-industrial control and evaluated on a separate 400-year segment (model years 0001-0400) that was never used in training. The only fitted object in the prediction chain is the emulator itself, and its performance is assessed out-of-sample, so no fitted parameter has been renamed as a prediction. The 0.09 W/m2 energy correction is imposed from E3SMv3's diagnosed residual imbalance rather than claimed as an emulator output. The stochastic-versus-deterministic comparison is an empirical controlled experiment with two seeds each, and the paper honestly reports residual gaps in Gulf Stream SST variance and MIZ sea-ice variance; these are limitations, not circular steps. Citations to Duncan et al. (2026), Perkins et al. (2026), and Dheeshjith et al. (2025) supply component architectures and training recipes, but the headline evaluation of E3SMv3 fidelity is produced by the paper's own held-out tests and does not reduce to those citations. The structural statement that ocean variability is inherited from ACE2S follows from the design choice of injecting no noise into Samudra, but the quantitative claim that the stochastic system recovers more variability than the deterministic baseline is measured, not assumed. No self-definitional, fitted-input, self-citation-load-bearing, imported-uniqueness, or ansatz-smuggling pattern is present.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced. The load-bearing assumptions are about stationarity, coarse-grid sufficiency, the source of stochasticity, and several training choices made without sensitivity analysis. The only hand-picked numerical constant that directly influences the coupled physics is the 0.09 W/m2 energy correction, taken from E3SMv3 itself.

free parameters (5)
  • Total energy budget correction constant = 0.09 W/m2
    Imposed as a constant unaccounted heating term in ACE2S fine-tuning, taken from E3SMv3's residual surface minus TOA energy imbalance (Section 2.2.2). It is an input, not a fitted output, but it is a hand-chosen constant affecting long-run drift.
  • Ocean vertical level depths = 19 levels from 10 m to 5500 m
    A modified set of levels chosen to align with E3SMv3's native vertical coordinate (Section 2.2.1); affects vertical interpolation and ocean state representation.
  • Stochastic loss weights and rollout sampling probabilities = CRPS 0.9, energy 0.1; per-batch windows
    Hyperparameters chosen without reported sensitivity analysis in Sections 2.2.4 and Text S1; these shape the probabilistic objective and the coupled training dynamics.
  • Ensemble size = 2 members
    All stochastic losses and rollouts use two ensemble members (Text S1). The authors note they did not observe the small-ensemble degeneracy associated with fair CRPS.
  • Spectral land-fill diagnostic parameters = n=4, k=5, sigma=1 grid cell
    Used only for ocean spectral diagnostics (Text S2), not for training; included for completeness.
assumptions (5)
  • domain assumption Stationarity of the E3SMv3 pre-industrial control simulation
    Training on years 0401-0505 and evaluating on years 0001-0400 of the same control run presumes the climate is stationary, so the two segments are exchangeable under the same distribution (Section 2.1).
  • domain assumption Coarse-grained 1-degree fields contain sufficient information to reproduce E3SMv3's coupled behavior
    The emulator operates at 1 degree horizontal resolution with 8 atmosphere and 19 ocean levels (Section 2.2); fidelity claims are limited to this coarse representation.
  • domain assumption Atmosphere-driven stochasticity suffices for ocean variability
    No noise is injected into Samudra during coupled fine-tuning; all stochasticity comes from ACE2S (Section 2.2.4). The authors flag this as an open question in Section 4.
  • ad hoc to paper Energy budget correction with a constant residual enforces conservation
    A vertically and horizontally uniform air temperature correction is applied in ACE2S fine-tuning (Section 2.2.2); it is a practical closure rather than a physically derived one.
  • ad hoc to paper Equal loss weighting across variables is appropriate
    The authors depart from previous per-variable loss weights without sensitivity analysis (Section 2.2.2), so the objective's balance is an assumption about what matters for downstream skill.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Stochastic Emulation of a Fully Coupled Preindustrial E3SMv3 Simulation." pith.science (2026). https://pith.science/paper/RSXMXEEP

@misc{pith2026260810277,
  author       = {Pith},
  title        = {Pith review of: Stochastic Emulation of a Fully Coupled Preindustrial E3SMv3 Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RSXMXEEP}},
  note         = {Machine review of arXiv:2608.10277}
}
read the original abstract

We present a stochastic coupled emulator of E3SM version 3, built on the SamudrACE framework, which couples an atmosphere emulator (ACE2) with a full-depth ocean emulator (Samudra). We replace the deterministic atmosphere emulator with its stochastic counterpart, ACE2S, and fine-tune the coupled system with a probabilistic objective, so that the atmosphere acts as a source of internal variability for the ocean. Trained on 105 years of a pre-industrial control simulation and evaluated on an independent 400 years, the emulator reproduces E3SMv3's mean climate state with biases much smaller than existing model-to-observation differences. Relative to a deterministic baseline, stochastic training maintains internal variability across timescales, most notably in the ENSO power spectrum, eddy-rich SST anomalies, and sea ice variability in the marginal ice zone. The emulator captures daily precipitation accurately up to the 99.99th percentile, but underestimates the rarest tropical extremes. These results show that stochastic coupled emulators can reproduce long-timescale variability with high fidelity, while extrapolation to unseen extremes remains a key challenge.

Figures

Figures reproduced from arXiv: 2608.10277 by the authors.

Figure 1
Figure 1. Overview of the SamudrACE-E3SMv3 training protocol. E3SMv3 achieves inference speeds of 1105 SYPD (stochastic) and 1360 SYPD (deter￾ministic). 3 Results 3.1 Climatology Consistent with the findings of Duncan et al. (2026), SamudrACE-E3SMv3 accu￾rately reproduces the spatial structure of the mean climate state of E3SMv3. We com￾pare the time-mean biases between SamudrACE-E3SMv3 and E3SMv3 against the ex￾isting biases… view at source ↗
Figure 2
Figure 2. 400-year time mean of surface temperature and precipitation for E3SMv3 (a and e), SamudrACE-E3SMv3 (b and f), bias between SamudrACE-E3SMv3 and E3SMv3 (c and g), and bias between E3SMv3 and observation (d and h). For comparison with HadISST, we define ocean grid cells as ocean fraction greater than 96% and only compare SST over these points. Note the observation datasets use different time spans that are specified i… view at source ↗
Figure 3
Figure 3. Taylor diagram of 400-year time mean atmosphere (a) and ocean (b) variables be￾tween SamudrACE-E3SMv3 and E3SMv3. Colors distinguish variables; shading darkens with increasing level index, from levels 0–7 in the atmosphere and 0–18 in the ocean. Level 0 is the lightest shade, corresponding to the top of the atmosphere in (a) and the near-surface layer in (b). Total water in (a) is specific total water, which is the … view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Daily precipitation probability density over the (a) global, (b) tropical ocean, and (c) CONUS domains, comparing E3SMv3 with SamudrACE-E3SMv3. Solid E3SMv3 lines show the 100-year training-period data; solid SamudrACE-E3SMv3 lines show the first 100 years of a free-ru…
Figure 5
Figure 5. Figure 5: ENSO characteristics in the 400-year E3SMv3 simulation and SamudrACE-E3SMv3 for (a) time series of monthly mean Ni˜no 3.4 index; (b) corresponding power spectra, where thin lines show the spectra of individual 40-year blocks and thick lines their mean; (c) regression o…
Figure 6
Figure 6. Figure 6: Sea surface temperature variability in the Gulf Stream Extension region (35◦ –45◦N, 75◦ –45◦W) over the 400-year evaluation. Comparison of the E3SMv3 target, the deterministic SamudrACE emulator, and the stochastic SamudrACE emulator for (a) power spectral den￾sity of …
Figure 7
Figure 7. Figure 7: Local sea-ice variability in the Northern (a-c) and Southern (d-f) Hemisphere: standard deviation of deseasonalized monthly sea-ice fraction anomalies over the 400-year eval￾uation period for the E3SMv3 target (a, d), the deterministic SamudrACE emulator (b, e), and th…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 18 canonical work pages

  1. [1]

    Duncan, James P. C. and Wu, Elynn and Dheeshjith, Surya and Subel, Adam and Arcomano, Troy and Clark, Spencer K. and Henn, Brian and Kwa, Anna and McGibbon, Jeremy and Perkins, W. Andre and Gregory, William and Fernandez-Granda, Carlos and Busecke, Julius and Watt-Meyer, Oliver and Hurlin, William J. and Adcroft, Alistair and Zanna, Laure and Bretherton, ...

    doi:10.48550/arxiv.2509.12490

  2. [2]

    doi:10.1029/2025GL119340 , number =

    Geophysical Research Letters , author =. doi:10.1029/2025GL119340 , number =

    doi:10.1029/2025gl119340

  3. [3]

    Geophysical Research Letters , author =

    Samudra:. Geophysical Research Letters , author =. doi:10.1029/2024GL114318 , abstract =

    doi:10.1029/2024gl114318

  4. [4]

    Geophysical Research Letters , author =

    Diversity of. Geophysical Research Letters , author =. 2018 , pages =. doi:10.1029/2018GL079203 , language =

    doi:10.1029/2018gl079203 2018

  5. [5]

    Journal of Advances in Modeling Earth Systems , author =

    The. Journal of Advances in Modeling Earth Systems , author =. doi:10.1029/2025MS005302 , number =

    work page doi:10.1029/2025ms005302

  6. [6]

    and Hack, James J

    Hurrell, James W. and Hack, James J. and Shea, Dennis and Caron, Julie M. and Rosinski, James , year =. A. Journal of Climate , publisher =. doi:10.1175/2008JCLI2292.1 , abstract =

    work page doi:10.1175/2008jcli2292.1

  7. [7]

    and Sapiano, Mathew R

    Adler, Robert F. and Sapiano, Mathew R. P. and Huffman, George J. and Wang, Jian-Jian and Gu, Guojun and Bolvin, David and Chiu, Long and Schneider, Udo and Becker, Andreas and Nelkin, Eric and Xie, Pingping and Ferraro, Ralph and Shin, Dong-Bin , year =. The. Atmosphere , publisher =. doi:10.3390/atmos9040138 , abstract =

    doi:10.3390/atmos9040138

  8. [8]

    2025 , pages =

    npj Climate and Atmospheric Science , author =. 2025 , pages =. doi:10.1038/s41612-025-01090-0 , language =

    doi:10.1038/s41612-025-01090-0 2025

Show all 41 references
  1. [9]

    and Hurlin, Bill and Watt-Meyer, Oliver and Adcroft, Alistair and Bretherton, Chris and Zanna, Laure , year =

    Gregory, William and Bushuk, Mitchell and Duncan, James and Wu, Elynn and Subel, Adam and Clark, Spencer K. and Hurlin, Bill and Watt-Meyer, Oliver and Adcroft, Alistair and Bretherton, Chris and Zanna, Laure , year =. doi:10.48550/arXiv.2603.12449 , publisher =

    doi arXiv

  2. [10]

    arXiv preprint arXiv:2504.06007 , year=

    CAMulator: Fast emulation of the community atmosphere model , author=. arXiv preprint arXiv:2504.06007 , year=

    arXiv

  3. [11]

    doi:10.1029/2025JH000686 , year =

    Aouni, Anass El and Gaudel, Quentin and Regnier, Charly and Gennip, Simon Van and Galloudec, Olivier Le and Drevillon, Marie and Drillet, Yann and Lellouche, Jean-Michel , journal=. doi:10.1029/2025JH000686 , year =

    doi

  4. [12]

    arXiv preprint arXiv:2308.03152 , year=

    Ai-goms: Large ai-driven global ocean modeling system , author=. arXiv preprint arXiv:2308.03152 , year=

    arXiv

  5. [13]

    arXiv preprint arXiv:2402.02995 , year=

    Xihe: A data-driven model for global ocean eddy-resolving forecasting , author=. arXiv preprint arXiv:2402.02995 , year=

    arXiv

  6. [14]

    and Williamson, Daniel and Challenor, Peter , year =

    Baker, Evan and Harper, Anna B. and Williamson, Daniel and Challenor, Peter , year =. Emulation of high-resolution land surface models using sparse. doi:10.5194/gmd-15-1913-2022 , journal =

    1913 doi

  7. [15]

    and Pappenberger, Florian and Balsamo, Gianpaolo , year =

    Wesselkamp, Marieke and Chantry, Matthew and Pinnington, Ewan and Choulga, Margarita and Boussetta, Souhail and Kalweit, Maria and Bödecker, Joschka and Dormann, Carsten F. and Pappenberger, Florian and Balsamo, Gianpaolo , year =. Advances in land surface forecasting: a compa...

    doi

  8. [16]

    Data-driven surrogate modeling of high-resolution sea-ice thickness in the

    Durand, Charlotte and Finn, Tobias Sebastian and Farchi, Alban and Bocquet, Marc and Boutin, Guillaume and Ólason, Einar , year =. Data-driven surrogate modeling of high-resolution sea-ice thickness in the. doi:10.5194/tc-18-1791-2024 , journal =

    doi

  9. [17]

    Generative

    Finn, Tobias Sebastian and Bocquet, Marc and Rampal, Pierre and Durand, Charlotte and Porro, Flavia and Farchi, Alban and Carrassi, Alberto , year =. Generative. doi:10.48550/arXiv.2508.14984 , publisher =

    doi

  10. [18]

    and Senior, Catherine A

    Eyring, Veronika and Bony, Sandrine and Meehl, Gerald A. and Senior, Catherine A. and Stevens, Bjorn and Stouffer, Ronald J. and Taylor, Karl E. , year =. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization , volume =. Geo...

    doi

  11. [19]

    Geoscientific Model Development , author =

    An evolving. Geoscientific Model Development , author =. 2025 , pages =. doi:10.5194/gmd-18-6671-2025 , number =

    2025 doi

  12. [20]

    Nature , volume=

    Neural general circulation models for weather and climate , author=. Nature , volume=. 2024 , publisher=

    2024

  13. [21]

    and Liu, Zihui and Espinosa, Zachary I

    Cresswell-Clay, Nathaniel and Liu, Bowen and Durran, Dale R. and Liu, Zihui and Espinosa, Zachary I. and Moreno, Raul A. and Karlbauer, Matthias , year =. A. doi:10.1029/2025AV001706 , journal =

    doi

  14. [22]

    and Watt-Meyer, Oliver and Kwa, Anna and McGibbon, Jeremy and Henn, Brian and Perkins, W

    Clark, Spencer K. and Watt-Meyer, Oliver and Kwa, Anna and McGibbon, Jeremy and Henn, Brian and Perkins, W. Andre and Wu, Elynn and Harris, Lucas M. and Bretherton, Christopher S. , year =. doi:10.1029/2024JH000575 , journal =

    doi

  15. [23]

    arXiv preprint arXiv:2512.18224 , year=

    HiRO-ACE: Fast and skillful AI emulation and downscaling trained on a 3 km global storm-resolving model , author=. arXiv preprint arXiv:2512.18224 , year=

  16. [24]

    Nature , author =

    Accurate medium-range global weather forecasting with 3D neural networks , DOI =. Nature , author =

  17. [25]

    Science , author =

    Learning skillful medium-range global weather forecasting , DOI =. Science , author =

  18. [26]

    Nature Communications , author =

    A machine learning model that outperforms conventional global subseasonal forecast models , DOI =. Nature Communications , author =

  19. [27]

    Neural general circulation models for weather and climate , DOI =

    Kochkov, Dmitrii and Yuval, Janni and Langmore, Ian and Norgaard, Peter and Smith, Jamie and Mooers, Griffin and Kl\". Neural general circulation models for weather and climate , DOI =

  20. [28]

    Duncan, James P. C. and Wu, Elynn and Golaz, Jean-Christophe and Caldwell, Peter M. and Watt-Meyer, Oliver and Clark, Spencer K. and McGibbon, Jeremy and Dresdner, Gideon and Kashinath, Karthik and Bonev, Boris and Pritchard, Michael S. and Bretherton, Christopher S. , year =....

    doi

  21. [29]

    , year =

    Wu, Elynn and Rebassoo, Finn and Paul, Pappu and Proistosescu, Cristian and Nugent, Jacqueline and McCoy, Daniel and Caldwell, Peter and Bretherton, Christopher S. , year =. Applying the. doi:10.1029/2025JH000774 , journal =

    doi

  22. [30]

    Qiang and Hassanzadeh, Pedram and Zand, Mohsen and Chattopadhyay, Ashesh and Weare, Jonathan and Abbot, Dorian S

    Sun, Y. Qiang and Hassanzadeh, Pedram and Zand, Mohsen and Chattopadhyay, Ashesh and Weare, Jonathan and Abbot, Dorian S. , year =. Can. doi:10.1073/pnas.2420914122 , journal =

    doi

  23. [31]

    Journal of Geophysical Research: Atmospheres , author =

    Summarizing multiple aspects of model performance in a single diagram , doi =. Journal of Geophysical Research: Atmospheres , author =

  24. [32]

    Autocalibration of the

    Yarger, Drew and Wagman, Benjamin Moore and Chowdhary, Kenny and Shand, Lyndsay , year =. Autocalibration of the. doi:10.1029/2023MS003961 , journal =

    doi

  25. [33]

    Climate dynamics , volume=

    A tripole index for the interdecadal Pacific oscillation , author=. Climate dynamics , volume=. 2015 , publisher=

    2015

  26. [34]

    Andre and Kwa, Anna and McGibbon, Jeremy and Arcomano, Troy and Clark, Spencer K

    Perkins, W. Andre and Kwa, Anna and McGibbon, Jeremy and Arcomano, Troy and Clark, Spencer K. and Watt-Meyer, Oliver and Bretherton, Christopher S. and Harris, Lucas M. , year =. doi:10.48550/arXiv.2512.18224 , publisher =

    doi

  27. [35]

    Samudra 2:

    Yuan, Yuan and Rusak, Jesse and Merose, Alexander and Subel, Adam and Perezhogin, Pavel and Adcroft, Alistair and Fernandez-Granda, Carlos and Zanna, Laure , year =. Samudra 2:. doi:10.48550/arXiv.2606.02610 , publisher =

    doi arXiv

  28. [36]

    and Arcomano, Troy and Duncan, James P

    Clark, Spencer K. and Arcomano, Troy and Duncan, James P. C. and Henn, Brian and Kwa, Anna and McGibbon, Jeremy and Perkins, W. Andre and Wu, Elynn and Harris, Lucas M. and Watt-Meyer, Oliver and Bretherton, Christopher S. , year =. Disentangling the effects of sea surface tem...

    doi arXiv

  29. [37]

    and Koldunov, Nikolay and Lessig, Christian and Molina, Maria J

    Henn, Brian and Bretherton, Christopher S. and Koldunov, Nikolay and Lessig, Christian and Molina, Maria J. and Arcomano, Troy and Watt-Meyer, Oliver and Couairon, Guillaume and Singh, Renu and Brunstein, Robert and Hasson, Yana and Jost, Antonia and Brenowitz, Noah and Mansha...

    doi arXiv

  30. [38]

    Geophysical Research Letters , year =

    Are historical records sufficient to constrain. Geophysical Research Letters , year =. doi:10.1029/2009GL038710 , author =

    doi

  31. [39]

    npj Artificial Intelligence , volume=

    Aifs-crps: ensemble forecasting using a model trained with a loss function based on the continuous ranked probability score , author=. npj Artificial Intelligence , volume=. 2026 , publisher=

    2026

  32. [40]

    Journal of the American statistical Association , volume=

    Strictly proper scoring rules, prediction, and estimation , author=. Journal of the American statistical Association , volume=. 2007 , publisher=

    2007

  33. [41]

    McGibbon, Jeremy and Watt-Meyer, Oliver and Duncan, James and Kwa, Anna and Henn, Brian and Clark, Spencer and Wu, Elynn and Perkins, W. Andre and Arcomano, Troy and Dodson, Anna and rebassoo and Mahfouz, Naser and Yermakov, Alexey and Yik, William and Dheeshjith, Surya and Gr...

    2026 doi

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.