Pith. sign in

REVIEW 3 major objections 5 minor 38 references

LanTu: Dynamics-Enhanced Deep Learning for Eddy-Resolving Ocean Forecasting

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A regional AI ocean model trained with an eddy-dynamics loss reports lower forecast errors than numerical and AI operational systems for lead times beyond 10 days.

desk verdict LanTu shows convincing operational gains against independent observations, but the paper never isolates its claimed novelty, so the causal story isn't established. read the letter →

arxiv 2505.10191 v1 pith:XYMW2LY3 submitted 2025-05-15 physics.ao-ph cs.AIcs.LGnlin.CD

Qingyu Zheng , Qi Shao , Guijun Han , Wei Li , Hong Li , Xuan Wang This is my paper · ORCID
classification physics.ao-phcs.AIcs.LGnlin.CD
keywords eddy-resolvingoceanforecastingdeeplearningmesoscaleeddiesdynamics-enhancedmultiscaleconstraintnon-autoregressivesealevelanomalyregionalmodeling
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

LanTu is a regional AI ocean forecasting system for the upper ocean (0–643 m) that predicts temperature, salinity, currents, and sea level anomaly at 1/12° resolution over the Northwestern Pacific. The paper aims to establish that adding a dynamics-informed training constraint—one that rewards matching the spatial pattern of change between the initial state and the forecast—lets an AI model preserve mesoscale eddies instead of smoothing them away. The reported result is that LanTu beats operational numerical forecast systems and a global AI forecast system on all four variable types for lead times of at least 10 days, with RMSE reductions of 46.78% for temperature and 20.08% for salinity at 6 days. If this holds, it matters because eddy-resolving forecasts are costly for numerical models, and a lightweight AI system that keeps eddies sharp for two to four weeks would be a practical tool for fisheries, navigation, and regional ocean services.

What carries the argument

The load-bearing mechanism is the dynamics-enhanced multiscale constraint, a two-part loss: $$\mathrm{Loss} = \lambda_S L_S + \lambda_D L_D, \quad \lambda_S = \lambda_D = 0.5,$$ with $L_S$ the mean squared error between forecast and target state and $L_D = 1 - \mathrm{Corr}(\Delta \hat{O}, \Delta O)$, where $\Delta \hat{O}$ and $\Delta O$ are the forecast and ground-truth increments relative to the initial field. Minimizing $L_D$ maximizes the Pearson correlation between predicted and true change, so the model is forced to get the spatial pattern of evolution right, not just the magnitude. This is paired with a non-autoregressive strategy—one trained model per lead time—and an architecture that mixes Fourier spatial tokens with cross-variable linear layers so atmospheric driver fields and ocean variables interact. Together they counteract the smoothing from norm-based losses that otherwise erases mesoscale eddy signals.

What would settle it

Run LanTu's training pipeline with $L_D$ removed (static MSE loss only), keeping architecture, data region, resolution, and the one-model-per-lead-time scheme identical; if the eddy metrics and RMSE advantages largely persist, the paper's central mechanism is not responsible for them.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that LanTu, a non-autoregressive vision-transformer model trained separately for each forecast lead time, outperforms both the operational numerical ocean forecast systems represented in the Class 4 verification benchmark and the global AI forecast system on temperature, salinity, sea level anomaly, and current forecasts in its regional domain, with the advantage persisting beyond 10 days and up to 30 days. The quantitative headline is a 46.78% lower RMSE for temperature and 20.08% lower for salinity at 6-day lead time versus the numerical benchmark, and vertical profile errors at 10 days that are 35.93% and 14.81% lower. The paper further claims that LanTu captures mesoscale eddy splitting and merging events and 3D eddy structure that the global AI system blurs or misses, and that its non-autoregressive strategy avoids the cumulative error that makes the autoregressive variant unskillful beyond 10 days. On the paper's terms, these results establish dynamics-enhanced deep learning as a viable regional eddy-resolving forecasting paradigm.

Load-bearing premise

The claimed advantage of the new training loss over all alternative explanations is assumed, because the paper never runs the same model with that loss removed.

Editorial extensions

If this is right

  • At lead times of 10–30 days, LanTu reports forecasts of temperature, salinity, and sea level anomaly with anomaly correlation above 0.6 and persistence skill scores that stay positive, meaning the model remains informative for two to four weeks.
  • The paper's eddy case studies imply that a properly constrained AI model can reproduce eddy splitting and merging as discrete events rather than smoothed averages, which is what regional forecasting applications need.
  • Because LanTu is a single regional model trained once per lead time, it offers a computationally cheap alternative to eddy-resolving numerical systems for regions with limited supercomputing capacity.
  • The comparison with the autoregressive variant indicates that non-autoregressive, lead-time-specific training is what prevents error accumulation in this setup; autoregressive rollouts lose skill beyond 10 days.

Reading between the lines

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

  • A testable extension the paper leaves implicit: an ablation of the dynamic loss would isolate how much of the gain comes from the increment-correlation term versus from regional training, lead-time-specific models, or larger effective domain.
  • Because the same smoothing problem afflicts other learned spatiotemporal forecasters, the increment-correlation loss could transfer to coastal circulation, sea-ice, or weather nowcasting tasks where conserving small-scale features matters.
  • The paper's non-autoregressive advantage is partly relative to an unfine-tuned autoregressive baseline; error-correcting or fine-tuned autoregressive models might narrow that gap, so the claim should be read as comparing strategies under equal training effort.
  • If the loss's effect is confirmed, it suggests that RMSE alone understates forecast quality; pattern-of-change correlation is what preserves dynamically important eddy features.
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. The manuscript presents LanTu, a regional eddy-resolving ocean forecasting system for the Northwestern Pacific, based on a Vision Transformer / Adaptive Fourier Neural Operator architecture with a non-autoregressive forecasting strategy. The model is trained on GLORYS reanalysis and ERA5 atmospheric forcing. The claimed novelty is a 'dynamics-enhanced multiscale constraint' (Eqs. 2–4) that adds a Pearson-correlation loss on forecast increments to the standard MSE loss. The paper reports that LanTu outperforms operational numerical ocean forecasts (IV-TT Class 4 systems) in temperature, salinity, sea level anomaly, and currents at lead times up to 10 days, with RMSE reductions of 46.78% for temperature and 20.08% for salinity at 6 days. Beyond 10 days, LanTu is compared against persistence and an autoregressive variant (LTAR), and its eddy dynamics are compared against the global AI model XiHe using GLORYS as ground truth. The authors argue that the dynamics-enhanced constraint mitigates the blurring effect and improves eddy splitting/merging forecasts and 3D eddy structure.

Significance. If the results hold, LanTu would be a useful demonstration that regionally trained AI ocean forecasting systems can exceed global operational numerical forecasts in a limited domain, with substantial computational savings. The use of independent IV-TT observations and DUACS altimetry for the headline skill metrics is a genuine strength, as is the evaluation against an established operational intercomparison framework. The central causal claim, however—that the dynamics-enhanced multiscale constraint is what drives the skill improvement—is not isolated by any ablation experiment. Because the comparisons to LTAR and XiHe change multiple factors simultaneously, the paper currently does not establish that the proposed loss term is the source of the improvement. The scientific significance of the methodological contribution therefore remains unproven, even though the empirical forecast skill is credible.

major comments (3)
  1. [Section 4.2.4, Eqs. (2)–(4)] The paper's central claim—that the dynamics-enhanced multiscale constraint improves forecast skill—is not tested. No experiment trains the same architecture with and without the D_loss term while holding all other factors fixed. The LTAR baseline changes the forecasting strategy (autoregressive vs. non-autoregressive) and adds atmospheric outputs, and the XiHe comparison changes architecture and training domain. Thus the reported improvements cannot be causally attributed to the increment-correlation loss. An ablation that removes D_loss (or varies lambda_D) is required to support the abstract's and Section 3's attribution.
  2. [Section 2.2 and 2.3, Figs. 3–5] The eddy-dynamics comparisons use GLORYS reanalysis as ground truth, and GLORYS is also the training target for both LanTu and XiHe. A model trained on a particular reanalysis can be closer to that reanalysis than another model without being a better predictor of the real ocean. While this does not favor LanTu over XiHe in an obvious one-sided way, it does mean that the claimed superiority in capturing eddy splitting, merging, and 3D structure is only a measure of closeness to the training product, not to independent ocean state. An independent eddy census (e.g., from altimetry or Argo-based eddy detection) would strengthen the eddy-dynamics claim. As it stands, the only fully independent evidence for eddy skill is the DUACS SLA comparison in Fig. 1A and Fig. S4, which is limited to surface patterns.
  3. [Abstract and Section 2.1] The abstract states that LanTu outperforms NOFS and AI-OFS 'with a lead time of more than 10 days.' The direct comparison to IV-TT numerical forecasts is limited to lead times of 1–10 days (Fig. 1, Fig. 2, Fig. S1). Beyond 10 days, the only benchmarks are persistence and the autoregressive LTAR variant (Figs. S2–S3); no same-lead-time head-to-head with an operational NOFS is shown at 11–30 days. The comparison in Fig. S1 of LanTu-30 against IV-TT-10 is a cross-lead-time comparison and does not support the phrasing that the outperformance over NOFS extends beyond 10 days. The claim should be reworded to distinguish 'skillful beyond 10 days' from 'outperforming NOFS beyond 10 days.'
minor comments (5)
  1. [Supplementary Materials, Tables and Figs.] The term 'Hovmöller' is misspelled as 'Hofmöller' in the caption of Fig. S4 and in Table S2; the spelling should be made consistent.
  2. [Section 4.2.3, Eq. (1)] Equation (1) appears garbled in the manuscript text; the mathematical notation for the LanTu mapping is not rendered correctly and should be fixed.
  3. [Section 2.1, Fig. 2 caption] The caption states 'Forecast lead times of 1-7 days' but panels A–D and E–H appear to cover 1–7 days; the text should clarify whether the 10-day data shown in the figure or only in Fig. S1, as the caption is ambiguous.
  4. [Materials and Methods, Data availability] The manuscript states that inference code and weights for XiHe are available, but no code or weights are provided for LanTu. Given the paper's emphasis on a new model, releasing code/weights would improve reproducibility.
  5. [Section 2.1, 'PSS' paragraph] The sentence 'The PSS of LanTu is consistently greater than 0 and shows a positive trend over the 10-30 days lead time' would benefit from a precise definition of the averaging period used to compute each PSS point, since the figure shows a curve over lead time.

Circularity Check

1 steps flagged · score 2.0 of 10

Core forecast skill is verified against independent observations, but the eddy-dynamics evaluation uses GLORYS, the same reanalysis that supervises the model, as ground truth.

  1. fitted input called prediction [Section 2.2 (Figs. 3-5) and Materials and Methods 4.2.4, Eqs. (2)-(4)]
    "The dynamic constraint is the forecast increment correlation constraint based on Pearson correlation coefficient to better capture the mesoscale dynamic evolution... (A) Ground truth (GLORYS), (B) LanTu, (C) XiHe, showing the evolution of sea surface current and EKE... The EKE of LanTu is closer to the GLORYS reanalysis, with only a slight suppression in intensity."

    LanTu's dynamic loss (Eq. 4) explicitly maximizes the Pearson correlation between forecast increments and GLORYS increments, and the static loss (Eq. 3) regresses the full state to GLORYS. The eddy-dynamics evidence in Figs. 3-5 then uses GLORYS as 'ground truth' for EKE and 3D structure. Showing that LanTu's eddy evolution is closer to GLORYS than XiHe's is therefore partly a check of fit to the training-target product, not an independent test of dynamical fidelity. The 2021-2023 period is temporally out-of-sample, which mitigates the issue, but the product is the same reanalysis used for supervision, so this evaluation cannot independently establish the dynamics-enhanced constraint as the cause of the eddy improvement.

full rationale

The paper's headline forecast skill claims are not circular: LanTu is verified against IV-TT Class 4 observations (Argo, drifting buoys, satellite SLA/currents) that LanTu did not train on, against DUACS altimetry, and against persistence, and these comparisons support the core RMSE/ACC/PSS claims. The only identifiable reduction is in the eddy-dynamics section: the model is optimized to match GLORYS states and increments (Eqs. 2-4), and then its mesoscale eddy evolution (EKE, splitting/merging, 3D structure) is scored against GLORYS as 'ground truth.' That is a partial circularity in the specific claim that the dynamics constraint improves eddy forecasting; it demonstrates closeness to the training product, not independent physical skill. No load-bearing self-citation, imported uniqueness theorem, or ansatz-by-citation appears. The absence of a loss-ablation control is an experimental design gap rather than a circular derivation, so it does not raise the score beyond a mild 2.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central empirical claims rely on the accuracy of GLORYS as a training target and as a benchmark for eddy dynamics, plus the untested effectiveness of the increment-correlation loss. No new physical entities are introduced.

free parameters (1)
  • Loss weights lambda_S and lambda_D = 0.5 and 0.5
    The loss combines static MSE and dynamic increment-correlation terms with equal weights (Eq. 2). No sensitivity analysis or tuning procedure is reported, and the central improvement claim depends on this balance.
assumptions (3)
  • domain assumption GLORYS reanalysis is an accurate representation of the true ocean state and is suitable as both training target and evaluation ground truth.
    The paper trains LanTu on GLORYS and then uses GLORYS as the reference for eddy dynamics and 3D structure comparisons in Figs. 3-5. If GLORYS is biased, the skill estimates are biased.
  • ad hoc to paper Pearson correlation of forecast increments is a valid proxy for mesoscale eddy dynamics and improves forecast skill.
    The dynamic constraint (Eq. 4) is introduced without a derivation or ablation showing that optimizing this correlation translates to better eddy forecasts. The claim is plausible but unverified.
  • domain assumption Daily mean ERA5 atmospheric variables (SLP, T2M, U10, V10) are sufficient as atmospheric drivers for the regional forecast model.
    The model uses only daily-averaged surface forcing, ignoring sub-daily variability and radiative fluxes. The adequacy of this simplification is not tested.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LanTu: Dynamics-Enhanced Deep Learning for Eddy-Resolving Ocean Forecasting." pith.science (2026). https://pith.science/paper/XYMW2LY3

@misc{pith2026250510191,
  author       = {Pith},
  title        = {Pith review of: LanTu: Dynamics-Enhanced Deep Learning for Eddy-Resolving Ocean Forecasting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XYMW2LY3}},
  note         = {Machine review of arXiv:2505.10191}
}
read the original abstract

Mesoscale eddies dominate the spatiotemporal multiscale variability of the ocean, and their impact on the energy cascade of the global ocean cannot be ignored. Eddy-resolving ocean forecasting is providing more reliable protection for fisheries and navigational safety, but also presents significant scientific challenges and high computational costs for traditional numerical models. Artificial intelligence (AI)-based weather and ocean forecasting systems are becoming powerful tools that balance forecast performance with computational efficiency. However, the complex multiscale features in the ocean dynamical system make AI models still face many challenges in mesoscale eddy forecasting (especially regional modelling). Here, we develop LanTu, a regional eddy-resolving ocean forecasting system based on dynamics-enhanced deep learning. We incorporate cross-scale interactions into LanTu and construct multiscale physical constraint for optimising LanTu guided by knowledge of eddy dynamics in order to improve the forecasting skill of LanTu for mesoscale evolution. The results show that LanTu outperforms the existing advanced operational numerical ocean forecasting system (NOFS) and AI-based ocean forecasting system (AI-OFS) in temperature, salinity, sea level anomaly and current prediction, with a lead time of more than 10 days. Our study highlights that dynamics-enhanced deep learning (LanTu) can be a powerful paradigm for eddy-resolving ocean forecasting.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 32 canonical work pages

  1. [1]

    F. S. Paolo, D. Kroodsma, J. Raynor, T. Hochberg, P. Davis, J. Cleary, L. Marsaglia, S. Orofino, C. Thomas, P. Halpin, Satellite mapping reveals extensive industrial activity at sea. Nature 625, 85–91 (2024)

    work page 2024

  2. [2]

    Visbeck, Ocean science research is key for a sustainable future

    M. Visbeck, Ocean science research is key for a sustainable future. Nature Communications 9, 690 (2018)

    work page 2018

  3. [3]

    H. Xiao, S. Bao, J. Ren, Z. Xu, S. Xue, J. Liu, Global transboundary synergies and trade-offs among Sustainable Development Goals from an integrated sustainability perspective. Nature Communications 15, 500 (2024). MANUSCRIPT 19

    work page 2024

  4. [4]

    Izaguirre, I

    C. Izaguirre, I. J. Losada, P. Camus, J. L. Vigh, V. Stenek, Climate change risk to global port operations. Nature Climate Change 11, 14–20 (2021)

    work page 2021

  5. [5]

    B. S. Halpern, M. Frazier, J. Afflerbach, J. S. Lowndes, F. Micheli, C. O’Hara, C. Scarborough, K. A. Selkoe, Recent pace of change in human impact on the world’s ocean. Scientific Reports 9, 11609 (2019)

    work page 2019

  6. [6]

    B. S. Halpern, M. Frazier, J. Potapenko, K. S. Casey, K. Koenig, C. Longo, J. S. Lowndes, R. C. Rockwood, E. R. Selig, K. A. Selkoe, S. Walbridge, Spatial and temporal changes in cumulative human impacts on the world’s ocean. Nature Communications 6, 7615 (2015)

    work page 2015

  7. [7]

    S. A. Henson, C. Beaulieu, T. Ilyina, J. G. John, M. Long, R. Séférian, J. Tjiputra, J. L. Sarmiento, Rapid emergence of climate change in environmental drivers of marine ecosystems. Nature Communications 8, 14682 (2017)

    work page 2017

  8. [8]

    B. A. Storer, M. Buzzicotti, H. Khatri, S. M. Griffies, H. Aluie, Global cascade of kinetic energy in the ocean and the atmospheric imprint. Science Advances 9, eadi7420

    work page

Show all 38 references
  1. [9]

    Zhang, W

    Z. Zhang, W. Wang, B. Qiu, Oceanic mass transport by mesoscale eddies. Science 345, 322–324 (2014)

    2014

  2. [10]

    X. Ma, X. Zhang, L. Wu, Z. Tang, P. Yang, F. Song, Z. Jing, H. Chen, Y. Qu, M. Yuan, Z. Chen, B. Gan, Midlatitude mesoscale thermal Air-sea interaction enhanced by greenhouse warming. Nature Communications 15, 7699 (2024)

    2024

  3. [11]

    J.-Y. Lin, H. Ho, G. Gopalakrishnan, Z.-W. Zheng, R.-S. Tseng, J. Pan, C.-R. Ho, Q. Zheng, Typhoon induced mesoscale cyclonic eddy a long neglected linkage between atmosphere ocean and climate. npj Climate and Atmospheric Science 8, 64 (2025)

    2025

  4. [12]

    Y. Qu, S. Wang, Z. Jing, H. Wang, L. Wu, Spatial Structure of Vertical Motions and Associated Heat Flux Induced by Mesoscale Eddies in the Upper Kuroshio-Oyashio Extension. Journal of Geophysical Research: Oceans 127, e2022JC018781 (2022)

    2022

  5. [13]

    Groeskamp, J

    S. Groeskamp, J. H. LaCasce, T. J. McDougall, M. Rogé, Full-Depth Global Estimates of Ocean Mesoscale Eddy Mixing From Observations and Theory. Geophysical Research Letters 47, e2020GL089425 (2020)

    2020

  6. [14]

    A. A. Aslam, J. Schwendike, S. C. Peatman, A. J. Matthews, C. E. Birch, M. A. Bollasina, P. A. Barrett, M. V. C. Azaneu, Atmospheric Response to Mesoscale Ocean Eddies in the Maritime Continent. Journal of Geophysical Research: Atmospheres 130, e2024JD042606 (2025)

    2025

  7. [15]

    H. Seo, L. W. O’Neill, M. A. Bourassa, A. Czaja, K. Drushka, J. B. Edson, B. Fox-Kemper, I. Frenger, S. T. Gille, B. P. Kirtman, S. Minobe, A. G. Pendergrass, L. Renault, M. J. Roberts, N. Schneider, R. J. Small, A. Stoffelen, Q. Wang, Ocean Mesoscale and Frontal-Scale Ocean–A...

    2023

  8. [16]

    E. P. Chassignet, S. G. Yeager, B. Fox-Kemper, A. Bozec, F. Castruccio, G. Danabasoglu, C. Horvat, W. M. Kim, N. Koldunov, Y. Li, P. Lin, H. Liu, D. V. Sein, D. Sidorenko, Q. Wang, X. Xu, Impact of horizontal resolution on global ocean–sea ice model simulations based on the ex...

    2020

  9. [17]

    L. Y. Wan, Y. Liu, T. J. Ling, Development of a global high-resolution marine dynamic environmental forecasting system. Atmospheric and Oceanic Science Letters 11, 379–387 (2018)

    2018

  10. [18]

    Zanna, J

    L. Zanna, J. M. Brankart, M. Huber, S. Leroux, T. Penduff, P. D. Williams, Uncertainty and scale interactions in ocean ensembles: From seasonal forecasts to multidecadal climate predictions. Quarterly Journal of the Royal Meteorological Society 145, 160–175 (2019)

    2019

  11. [19]

    L. Chen, X. Zhong, F. Zhang, Y. Cheng, Y. Xu, Y. Qi, H. Li, FuXi: a cascade machine learning forecasting system for 15-day global weather forecast. npj Climate and Atmospheric Science 6, 190 (2023)

    2023

  12. [20]

    K. Bi, L. Xie, H. Zhang, X. Chen, X. Gu, Q. Tian, Accurate medium-range global weather forecasting with 3D neural networks. Nature 619, 533–538 (2023)

    2023

  13. [21]

    K. Chen, T. Han, J. Gong, L. Bai, F. Ling, J.-J. Luo, X. Chen, L. Ma, T. Zhang, R. Su, Y. Ci, B. Li, X. Yang, W. Ouyang, FengWu: Pushing the Skillful Global Medium-range Weather Forecast beyond 10 Days Lead (2023). https://arxiv.org/abs/2304.02948

    2023

  14. [22]

    Pathak, S

    J. Pathak, S. Subramanian, P. Harrington, S. Raja, A. Chattopadhyay, M. Mardani, T. Kurth, D. Hall, Z. Li, K. Azizzadenesheli, P. Hassanzadeh, K. Kashinath, A. Anandkumar, FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators (...

    2022 arXiv

  15. [23]

    R. Lam, A. Sanchez-Gonzalez, M. Willson, P. Wirnsberger, M. Fortunato, F. Alet, S. Ravuri, T. Ewalds, Z. Eaton-Rosen, W. Hu, A. Merose, S. Hoyer, G. Holland, O. Vinyals, J. Stott, A. Pritzel, S. Mohamed, P. Battaglia, Learning skillful medium-range global weather forecasting. ...

    2023

  16. [24]

    Xiong, Y

    W. Xiong, Y. Xiang, H. Wu, S. Zhou, Y. Sun, M. Ma, X. Huang, AI-GOMS: Large AI-Driven Global Ocean Modeling System (2023). https://arxiv.org/abs/2308.03152

    2023 arXiv

  17. [25]

    X. Wang, R. Wang, N. Hu, P. Wang, P. Huo, G. Wang, H. Wang, S. Wang, J. Zhu, J. Xu, J. Yin, S. Bao, C. Luo, Z. Zu, Y. Han, W. Zhang, K. Ren, K. Deng, J. Song, XiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting (2024). https://arxiv.org/abs/2402.02995

    2024 arXiv

  18. [26]

    N. Yang, C. Wang, M. Zhao, Z. Zhao, H. Zheng, B. Zhang, J. Wang, X. Li, LangYa: MANUSCRIPT 21 Revolutionizing Cross-Spatiotemporal Ocean Forecasting. arXiv arXiv:2412.18097 [Preprint] (2025). https://doi.org/10.48550/arXiv.2412.18097

    2025 doi arXiv

  19. [27]

    Y. Cui, R. Wu, X. Zhang, Z. Zhu, B. Liu, J. Shi, J. Chen, H. Liu, S. Zhou, L. Su, Z. Jing, H. An, L. Wu, Forecasting the eddying ocean with a deep neural network. Nature Communications 16, 2268 (2025)

    2025

  20. [28]

    A. G. Ryan, C. Regnier, P. Divakaran, T. Spindler, A. Mehra, G. C. Smith, F. Davidson, F. Hernandez, J. Maksymczuk, Y. L. and, GODAE OceanView Class 4 forecast verification framework: global ocean inter-comparison. Journal of Operational Oceanography 8, s98–s111 (2015)

    2015

  21. [29]

    Guiavarc’h, J

    C. Guiavarc’h, J. Roberts-Jones, C. Harris, D. J. Lea, A. Ryan, I. Ascione, Assessment of ocean analysis and forecast from an atmosphere–ocean coupled data assimilation operational system. Ocean Science 15, 1307–1326 (2019)

    2019

  22. [30]

    P. A. Sandery, P. Sakov, Ocean forecasting of mesoscale features can deteriorate by increasing model resolution towards the submesoscale. Nature Communications 8, 1566 (2017)

    2017

  23. [31]

    Peiris, V

    V. Peiris, V. Roshchina, N. Sukhorukova, Artificial neural networks with uniform norm-based loss functions. Advances in Computational Mathematics 50, 31 (2024)

    2024

  24. [32]

    D. Dong, P. Brandt, P. Chang, F. Schütte, X. Yang, J. Yan, J. Zeng, Mesoscale Eddies in the Northwestern Pacific Ocean: Three-Dimensional Eddy Structures and Heat/Salt Transports. Journal of Geophysical Research: Oceans 122, 9795–9813 (2017)

    2017

  25. [33]

    F. Chu, Z. Si, X. Yan, Z. Liu, J. Yu, C. Pang, Physical structure and evolution of a cyclonic eddy in the Northern South China sea. Deep Sea Research Part I: Oceanographic Research Papers 189, 103876 (2022)

    2022

  26. [34]

    Storto, A

    A. Storto, A. Alvera-Azcárate, M. A. Balmaseda, A. Barth, M. Chevallier, F. Counillon, C. M. Domingues, M. Drevillon, Y. Drillet, G. Forget, G. Garric, K. Haines, F. Hernandez, D. Iovino, L. C. Jackson, J.-M. Lellouche, S. Masina, M. Mayer, P. R. Oke, S. G. Penny, K. A. Peters...

    2019

  27. [35]

    Jean-Michel, G

    L. Jean-Michel, G. Eric, B.-B. Romain, G. Gilles, M. Angélique, D. Marie, B. Clément, H. Mathieu, L. G. Olivier, R. Charly, C. Tony, T. Charles-Emmanuel, G. Florent, R. Giovanni, B. Mounir, D. Yann, L. T. Pierre-Yves, The Copernicus Global 1/12° Oceanic and Sea Ice GLORYS12 Re...

    2021

  28. [36]

    Hersbach, B

    H. Hersbach, B. Bell, P. Berrisford, S. Hirahara, A. Horányi, J. Muñoz-Sabater, J. Nicolas, C. Peubey, R. Radu, D. Schepers, A. Simmons, C. Soci, S. Abdalla, X. Abellan, G. Balsamo, P. Bechtold, G. Biavati, J. Bidlot, M. Bonavita, G. De Chiara, P. Dahlgren, D. Dee, M. Diamanta...

    2020

  29. [37]

    Guibas, M

    J. Guibas, M. Mardani, Z. Li, A. Tao, A. Anandkumar, B. Catanzaro, Adaptive Fourier Neural Operators: Efficient Token Mixers for Transformers. arXiv arXiv:2111.13587 [Preprint] (2022). https://doi.org/10.48550/arXiv.2111.13587

    2022 doi arXiv

  30. [38]

    Taburet, A

    G. Taburet, A. Sanchez-Roman, M. Ballarotta, M.-I. Pujol, J.-F. Legeais, F. Fournier, Y. Faugere, G. Dibarboure, DUACS DT2018: 25 years of reprocessed sea level altimetry products. Ocean Science 15, 1207–1224 (2019). Supplementary Materials 1 Supplementary Materials for LanTu:...

    2019

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.