Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Finetuning a Weather Foundation Model with Lightweight Decoders for Unseen Physical Processes

T0 review · 3 major / 4 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 weather foundation model's frozen latent space carries enough physical information to predict unseen hydrological variables with lightweight decoders, with accuracy tracking physical correlation.

desk verdict A solid lightweight extension of a frozen weather foundation model, with a cost-saving result that holds; the overreach is the claim that decoder accuracy proves the latent space encodes physical relationships, because the missing raw-input control leaves a simpler explanation standing. read the letter →

arxiv 2506.19088 v1 pith:NU72PENM submitted 2025-06-23 cs.LG

Fanny Lehmann , Firat Ozdemir , Benedikt Soja , Torsten Hoefler , Siddhartha Mishra , Sebastian Schemm This is my paper · ORCID
classification cs.LG
keywords weatherfoundationmodelslatentspacelightweightdecodershydrologicalvariablesprecipitationpredictionfine-tuningphysicalcorrelationsEarthsystemforecast
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

This paper tests whether a weather foundation model can be extended to physical variables it never saw during pretraining without retraining the network. The authors freeze Aurora's latent space and train one shallow MLP decoder per new surface variable, targeting hydrological fields such as precipitation, potential evaporation, runoff, and soil moisture. The decoder-based extension reaches strong skill on variables that are physically correlated with Aurora's pretraining variables, uses roughly half the training time and 35 percent less memory than full fine-tuning, and inherits autoregressive stability. The paper reads this as evidence that a foundation model's latent space stores statistical relationships between physical variables, and proposes extensibility to new variables as a quality metric for Earth-science foundation models.

What carries the argument

The central object is Aurora's frozen latent vector: after the Swin-transformer processor, the state has shape $(HW/P^2) \times 4 \times 2E$, where the four levels are three atmospheric latent levels plus a surface level, and $E=512$ is the embedding dimension. The paper adds one lightweight MLP decoder per new variable; each decoder has three layers of widths $E$, $E/2$, $E/2$ with ReLU activations, about 300,000 parameters, and maps the surface-level latent to a patch of the target field. The machinery works because gradients flow only through the decoders, so training cost and memory drop while the pretrained representation stays fixed.

What would settle it

Take a target variable that is physically independent of all variables Aurora was pretrained on, such as terrestrial water storage, and train the same MLP decoder on it: if the decoder reconstructs the variable with high skill, accuracy is not driven by encoded physical correlation; if skill collapses to noise, the correlation mechanism is supported.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the latent representation produced by Aurora's processor—the four-level pressure-latent state that normally feeds the surface and atmospheric decoders—already contains enough information to reconstruct surface variables outside Aurora's training set. A three-layer MLP of about 300,000 parameters, trained on a latitude-weighted MAE against ERA5 (and a merged observational precipitation dataset for precipitation), predicts 6-hour accumulations of precipitation, potential evaporation, runoff, and soil moisture; for potential evaporation and soil moisture the correlation with the reference exceeds 0.95, for precipitation the fraction skill score reaches 0.92 at a 1 mm threshold and 0.82 at 5 mm, while runoff remains weaker. The same architecture also predicts eight surface and top-of-atmosphere energy fluxes with RMSE close to a task-specific model for several variables. The authors interpret the ordering of these accuracies as evidence that the latent space has encoded the physical coupling between variables: precipitation is learnable because it is tied to moisture-flux convergence of variables Aurora was trained on, while terrestrial water storage is not learnable because it depends on land-surface processes outside that set.

Load-bearing premise

The load-bearing premise is that Aurora's frozen latent representation already carries enough information about a new variable for a shallow MLP to reconstruct it; the paper demonstrates this for several variables but does not prove it holds generally.

Editorial extensions

If this is right

  • Hydrological variables absent from pretraining—precipitation, potential evaporation, runoff, soil moisture—can be forecast at 6-hour lead time from a frozen foundation model, with skill that approaches full fine-tuning for variables tightly coupled to the pretrained inputs.
  • Precipitation skill remains competitive with task-specific AI weather models and a numerical ensemble, and the decoders inherit stable autoregressive rollouts long enough to anticipate a Mediterranean cyclone's heavy rain several days ahead.
  • The cost of adding a variable drops to roughly half the training time and 65 percent of the GPU memory of full fine-tuning, because no gradients pass through the 1.3-billion-parameter backbone.
  • Decoder accuracy becomes a diagnostic: variables with weak physical ties to pretraining, such as terrestrial water storage, expose the boundary of what the latent space encodes.
  • For Earth-science foundation models, the paper proposes extensibility to new variables without fine-tuning as a quality metric distinct from downstream task accuracy.

Reading between the lines

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

  • A consequence left implicit is that the same protocol could be used to rank how well a foundation model understands different physical couplings, effectively turning the decoder into a physical-knowledge probe.
  • A cheap testable extension would be to train decoders on an interpolated family of targets, such as weighted mixtures of two variables, and verify that skill is monotone in the mixture coefficient.
  • The success of the frozen-latent approach suggests future pretraining should include a wider diversity of physical variables, since broader couplings in the latent space would lower the cost of downstream adaptation for resource-limited users.
  • For operational use, decoder predictions could be combined with a short fine-tune of only the last layers, rather than all parameters, to recover slow soil-moisture dynamics without full retraining.
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 / 4 minor

Summary. The paper investigates whether the frozen latent representation of the Aurora weather foundation model can support prediction of surface hydrological and energy variables that were not seen during pretraining. The authors train lightweight MLP decoders on Aurora's latent output, compare them against full fine-tuning of Aurora (Aurora+) and against GraphCast, FuXi, and IFS, and report that the decoders achieve good accuracy for many variables at roughly half the training time and 35% lower memory. The paper further argues that decoder accuracy scales with the physical correlation between the new variables and the variables used in Aurora's pretraining, which is interpreted as evidence that the latent space encodes meaningful physical relationships.

Significance. If the central interpretive claim were established, the paper would provide a practical and inexpensive way to extend weather foundation models to new variables and would introduce a useful evaluation criterion for foundation models in Earth science. The experimental setup is clearly specified, the code and datasets are openly available, and the evaluation against independent reanalysis products (ERA5, MSWEP) adds credibility to the reported skill. However, the main conceptual claim—that decoder accuracy reveals what Aurora's pretrained latent space encodes—is currently under-supported because the experiments lack a control for the pretrained encoder, do not quantify physical correlation, and provide no uncertainty estimates for the reported metrics.

major comments (3)
  1. [2.2, Fig. 2, Tables 1 and 3] The paper's strongest claim—that decoder accuracy indicates that Aurora's latent space 'captures meaningful physical relationships'—is not supported by the experiments as designed. The decoder is always applied to the output of the pretrained encoder-processor, so its predictions are a function of the raw input fields through Aurora. A decoder of similar capacity trained directly on the input fields, or on features from an untrained/randomly initialized encoder, could achieve comparable skill if the predictive relationships are statistical properties of the input data rather than something learned during pretraining. The manuscript does not include such a control, so the observed accuracy is also consistent with the weaker alternative that the pretrained model acts merely as a convenient feature extractor. Adding this ablation is necessary to justify the central claim.
  2. [Abstract, Sections 4.1-4.2 and 5] The statement that 'decoder accuracy depends on the physical correlation between the new variables and those used during pretraining' is nowhere quantified. The paper offers qualitative explanations for potential evaporation and precipitation and notes that runoff and terrestrial water storage are harder, but it never computes a numerical measure of correlation (e.g., pixelwise climatological correlation or mutual information between the target variable and the pretraining input variables at the relevant lead time). Without such a measure, the claimed dependence is not testable and could be confounded by variable-specific data difficulty or noise. The negative example of terrestrial water storage in Fig. B2 is presented without any metrics, which reduces the strength of the claimed negative result as well.
  3. [Tables 1, 3, and Fig. 7] All metrics are computed over a single test year (2020) with no uncertainty estimates. Weather fields are strongly autocorrelated in time, so the 1460 test samples are not independent. Confidence intervals from a block bootstrap or an equivalent procedure are needed to determine whether the differences across variables—for example, runoff PCC 0.42 versus precipitation PCC 0.71—are statistically meaningful. Without these, the qualitative ranking that underlies the paper's main claim remains vulnerable to noise.
minor comments (4)
  1. [Section 2.3 and Section 6] The paper states that forecasts at t+Δt do not rely on target variable observations at time t, but Section 2.3 describes Aurora+ as using the new variables as inputs. This apparent contradiction should be clarified, since it affects how the decoder and Aurora+ comparisons are interpreted.
  2. [Section 4.4, Fig. 7] The claim that the decoder yields RMSE close to ACE2 for several energy variables is based on raw RMSE values in different physical units, even though the figure caption correctly notes that errors should not be compared between variables. A normalized metric, such as RMSE relative to the climatological standard deviation of each variable, would make the comparison more informative.
  3. [Table 2] The abstract states '50% less training time,' but Table 2 reports sample throughput of 0.34 versus 0.16 samples per second, which corresponds to roughly a 53% reduction. The phrasing should be 'roughly half' or the exact ratio should be reported for consistency.
  4. [Section 4.2 and Table D1] When ERA5 is used as the precipitation reference, the decoder's advantage over GraphCast and FuXi disappears and its FSS values drop substantially; this is disclosed but should be emphasized more prominently in the main text to avoid the impression that the decoder is uniformly competitive with the benchmark models.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: decoder outputs are evaluated against external reanalysis data, and the physical-correlation claim is a post hoc interpretation rather than a fitted identity.

full rationale

The paper's derivation chain is self-contained and externally grounded. Aurora is a pretrained external model (Bodnar et al., 2025) whose weights are frozen; the only trainable components are small MLP decoders (~300k parameters) trained with a latitude-weighted MAE loss against ERA5/MSWEP targets over 2002-2014. The decoder outputs are then compared with held-out reanalysis data from 2020 (Tables 1 and 3), so the reported accuracies are measured against independent data, not recovered from the training objective. No fitted parameter is renamed as a prediction, and no target variable is defined in terms of the decoder output. The central claim that 'decoder accuracy depends on the physical correlation' is an empirical interpretation of observed skill differences across variables (e.g., potential evaporation vs. runoff), not an equation that reduces to the model's inputs. The paper even includes a negative case (terrestrial water storage) that is not well decoded, which is inconsistent with a claim that the latent space trivially contains all target variables. Self-citations (Herde et al. for corroboration of frozen-latent transfer; Gou & Soja as the TWS dataset source) are not load-bearing: the central evidence is the decoder evaluation itself, and neither citation defines the decoder predictions. The absence of a raw-input control decoder is a potential alternative explanation for the observed accuracy, but it does not make any step circular; it is a missing control, not a definitional identity. Hence no circularity.

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

No new physical entities are introduced. The central dependencies are the pretrained Aurora model, the ERA5/MSWEP data, and the interpretive claim that accuracy reflects physical relatedness.

assumptions (3)
  • domain assumption Aurora's frozen latent space is a sufficient representation of the target hydrological variables.
    The decoder approach relies entirely on the premise that the pretrained latent features retain the information needed to regress variables never seen during pretraining. Invoked in Section 2.2.
  • domain assumption ERA5 and MSWEP provide sufficiently accurate values for the target variables to serve as training targets and reference.
    All metrics and training losses depend on the fidelity of these reanalysis and precipitation datasets, described in Section 3.
  • ad hoc to paper The observed accuracy differences across variables reflect physical correlation with pretraining variables rather than differences in data difficulty or noise.
    The paper argues that accuracy follows physical relatedness, but it does not quantify correlation or control for data difficulty; this is the paper's interpretation of Table 1 and Section 5.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Finetuning a Weather Foundation Model with Lightweight Decoders for Unseen Physical Processes." pith.science (2026). https://pith.science/paper/NU72PENM

@misc{pith2026250619088,
  author       = {Pith},
  title        = {Pith review of: Finetuning a Weather Foundation Model with Lightweight Decoders for Unseen Physical Processes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NU72PENM}},
  note         = {Machine review of arXiv:2506.19088}
}
read the original abstract

Recent advances in AI weather forecasting have led to the emergence of so-called "foundation models", typically defined by expensive pretraining and minimal fine-tuning for downstream tasks. However, in the natural sciences, a desirable foundation model should also encode meaningful statistical relationships between the underlying physical variables. This study evaluates the performance of the state-of-the-art Aurora foundation model in predicting hydrological variables, which were not considered during pretraining. We introduce a lightweight approach using shallow decoders trained on the latent representations of the pretrained model to predict these new variables. As a baseline, we compare this to fine-tuning the full model, which allows further optimization of the latent space while incorporating new variables into both inputs and outputs. The decoder-based approach requires 50% less training time and 35% less memory, while achieving strong accuracy across various hydrological variables and preserving desirable properties of the foundation model, such as autoregressive stability. Notably, decoder accuracy depends on the physical correlation between the new variables and those used during pretraining, indicating that Aurora's latent space captures meaningful physical relationships. In this sense, we argue that an important quality metric for foundation models in Earth sciences is their ability to be extended to new variables without a full fine-tuning. This provides a new perspective for making foundation models more accessible to communities with limited computational resources, while supporting broader adoption in Earth sciences.

Figures

Figures reproduced from arXiv: 2506.19088 by the authors.

Figure 1
Figure 1. In black, the original Aurora model with its encoder-processor-decoder [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Next-step prediction at t+6h of a) total precipitation [mm/6hr], b) potential [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. (a) Average of precipitation prediction in Western Europe (area shown in [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: 24-hour accumulation of precipitation in Western Europe on October 2nd, [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Daily average of 6-hour time series of a) total precipitation [m/6hr], b) [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Relative RMSE computed on daily sums at the scale of a river basin. [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Root Mean Square Error (RMSE) of eight energy variables computed for lead [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Map of a) surface net solar radiation and b) surface net thermal radiation [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Data-driven solar forecasting enables near-optimal economic decisions

    physics.geo-ph 2025-09 conditional novelty 6.0 of 10

    SunCastNet combines AI weather forecasting with reinforcement-learning battery control to turn high-resolution solar forecasts into large regret reductions and more profitable industrial solar projects.

Reference graph

Works this paper leans on

28 extracted references · 8 canonical work pages · cited by 1 Pith paper

  1. [1]

    , Oskarsson, J

    adamov_building_2025 APACrefauthors Adamov, S. , Oskarsson, J. , Denby, L. , Landelius, T. , Hintz, K. , Christiansen, S. Schemm, S. APACrefauthors \ 2025 . Building Machine Learning Limited Area Models : Kilometer - Scale Weather Forecasting in Realistic Settings . Building Machine Learning Limited Area Models : Kilometer - Scale Weather Forecasting in R...

    arXiv doi:10.48550/arxiv.2504.09340 2025

  2. [2]

    , Wood, E F

    beck_mswep_2019 APACrefauthors Beck, H E. , Wood, E F. , Pan, M. , Fisher, C K. , Miralles, D G. , Van Dijk, A I J M. Adler, R F. APACrefauthors \ 2019 . MSWEP V2 Global 3- Hourly 0.1° Precipitation : Methodology and Quantitative Assessment MSWEP V2 Global 3- Hourly 0.1° Precipitation : Methodology and Quantitative Assessment . Bulletin of the American Me...

    doi:10.1175/bams-d-17-0138.1 2019

  3. [3]

    , Bruinsma, W P

    bodnar_foundation_2025 APACrefauthors Bodnar, C. , Bruinsma, W P. , Lucic, A. , Stanley, M. , Allen, A. , Brandstetter, J. Perdikaris, P. APACrefauthors \ 2025 . A foundation model for the Earth system A foundation model for the Earth system . Nature . APACrefDOI doi:10.1038/s41586-025-09005-y APACrefDOI

    doi:10.1038/s41586-025-09005-y 2025

  4. [4]

    , Hudson, D A

    bommasani_opportunities_2021 APACrefauthors Bommasani, R. , Hudson, D A. , Adeli, E. , Altman, R. , Arora, S. , von Arx, S. Liang, P. APACrefauthors \ 2021 . On the opportunities and risks of foundation models. On the opportunities and risks of foundation models. APACrefURL https://crfm.stanford.edu/assets/report.pdf APACrefURL

    work page 2021

  5. [5]

    , Zhong, X

    chen_fuxi_2023 APACrefauthors Chen, L. , Zhong, X. , Zhang, F. , Cheng, Y. , Xu, Y. , Qi, Y. \ Li, H. APACrefauthors \ 2023 . FuXi : a cascade machine learning forecasting system for 15-day global weather forecast FuXi : a cascade machine learning forecasting system for 15-day global weather forecast . npj Climate and Atmospheric Science 6 1 190 . APACref...

    doi:10.1038/s41612-023-00512-1 2023

  6. [6]

    , Leyland, J

    gebrechorkos_global-scale_2024 APACrefauthors Gebrechorkos, S H. , Leyland, J. , Dadson, S J. , Cohen, S. , Slater, L. , Wortmann, M. Darby, S E. APACrefauthors \ 2024 . Global-scale evaluation of precipitation datasets for hydrological modelling Global-scale evaluation of precipitation datasets for hydrological modelling . Hydrology and Earth System Scie...

    work page doi:10.5194/hess-28-3099-2024 2024

  7. [7]

    \ Soja, B

    gou_global_2024 APACrefauthors Gou, J. \ Soja, B. APACrefauthors \ 2024 . Global high-resolution total water storage anomalies from self-supervised data assimilation using deep learning algorithms Global high-resolution total water storage anomalies from self-supervised data assimilation using deep learning algorithms . Nature Water 2 2 139--150 . APACref...

    doi:10.1038/s44221-024-00194-w 2024

  8. [8]

    \ Finney, I

    hardy_leveraging_2025 APACrefauthors Hardy, L. \ Finney, I. APACrefauthors \ 2025 . Leveraging state‐of‐the‐art AI models to forecast wind power generation using deep learning Leveraging state‐of‐the‐art AI models to forecast wind power generation using deep learning . Meteorological Applications 32 2 e70038 . APACrefDOI doi:10.1002/met.70038 APACrefDOI

    work page doi:10.1002/met.70038 2025

Show all 28 references
  1. [9]

    , Raonić, B

    herde_poseidon_2024 APACrefauthors Herde, M. , Raonić, B. , Rohner, T. , Käppeli, R. , Molinaro, R. , de Bézenac, E. \ Mishra, S. APACrefauthors \ 2024 05 . Poseidon: Efficient Foundation Models for PDEs Poseidon: Efficient Foundation Models for PDEs . Advances in neural infor...

    2024

  2. [10]

    , Bell, B

    hersbach_era5_2020 APACrefauthors Hersbach, H. , Bell, B. , Berrisford, P. , Hirahara, S. , Horányi, A. , Muñoz-Sabater, J. Thépaut, J N. APACrefauthors \ 2020 . The ERA5 global reanalysis The ERA5 global reanalysis . Quarterly Journal of the Royal Meteorological Society 146 7...

    2020 doi

  3. [11]

    , Gimeno, F

    jaegle_perceiver_2021 APACrefauthors Jaegle, A. , Gimeno, F. , Brock, A. , Vinyals, O. , Zisserman, A. \ Carreira, J. APACrefauthors \ 2021 . Perceiver: General Perception with Iterative Attention Perceiver: General Perception with Iterative Attention . M. Meila\ T. Zhang\ ( )...

    2021

  4. [12]

    , Sanchez-Gonzalez, A

    lam_learning_2023 APACrefauthors Lam, R. , Sanchez-Gonzalez, A. , Willson, M. , Wirnsberger, P. , Fortunato, M. , Alet, F. Battaglia, P. APACrefauthors \ 2023 . Learning skillful medium-range global weather forecasting Learning skillful medium-range global weather forecasting ...

    2023 doi

  5. [13]

    \ Grill, G

    lehner_global_2013 APACrefauthors Lehner, B. \ Grill, G. APACrefauthors \ 2013 . Global river hydrography and network routing: baseline data and new approaches to study the world's large river systems Global river hydrography and network routing: baseline data and new approach...

    2013 doi

  6. [14]

    , Luise, I

    lessig_atmorep_2023 APACrefauthors Lessig, C. , Luise, I. , Gong, B. , Langguth, M. , Stadtler, S. \ Schultz, M. APACrefauthors \ 2023 . AtmoRep : A stochastic model of atmosphere dynamics using large scale representation learning. AtmoRep : A stochastic model of atmosphere dy...

    2023 doi arXiv

  7. [15]

    , Kusche, J

    li_forecasting_2024 APACrefauthors Li, F. , Kusche, J. , Sneeuw, N. , Siebert, S. , Gerdener, H. , Wang, Z. Tian, K. APACrefauthors \ 2024 . Forecasting Next Year 's Global Land Water Storage Using GRACE Data Forecasting Next Year 's Global Land Water Storage Using GRACE Data ...

    2024 doi

  8. [16]

    , Hsu, K

    liu_evaluation_2024 APACrefauthors Liu, C C. , Hsu, K. , Peng, M S. , Chen, D S. , Chang, P L. , Hsiao, L F. Kuo, H C. APACrefauthors \ 2024 . Evaluation of five global AI models for predicting weather in Eastern Asia and Western Pacific Evaluation of five global AI models for...

    2024 doi

  9. [17]

    , Brandstetter, J

    nguyen_climax_2023 APACrefauthors Nguyen, T. , Brandstetter, J. , Kapoor, A. , Gupta, J K. \ Grover, A. APACrefauthors \ 2023 . ClimaX : A foundation model for weather and climate ClimaX : A foundation model for weather and climate . A. Krause, E. Brunskill, K. Cho, B. Engelha...

    2023

  10. [18]

    , Ceola, S

    palazzoli_graice_2025 APACrefauthors Palazzoli, I. , Ceola, S. \ Gentine, P. APACrefauthors \ 2025 . GRAiCE : reconstructing terrestrial water storage anomalies with recurrent neural networks GRAiCE : reconstructing terrestrial water storage anomalies with recurrent neural net...

    2025 doi

  11. [19]

    , Garcia, N

    ramdas_wasserstein_2015 APACrefauthors Ramdas, A. , Garcia, N. \ Cuturi, M. APACrefauthors \ 2015 . On Wasserstein Two Sample Testing and Related Families of Nonparametric Tests . On Wasserstein Two Sample Testing and Related Families of Nonparametric Tests . arXiv . APACrefDO...

    2015 doi arXiv

  12. [20]

    , Dueben, P D

    rasp_weatherbench_2020 APACrefauthors Rasp, S. , Dueben, P D. , Scher, S. , Weyn, J A. , Mouatadid, S. \ Thuerey, N. APACrefauthors \ 2020 . WeatherBench : A Benchmark Data Set for Data ‐ Driven Weather Forecasting WeatherBench : A Benchmark Data Set for Data ‐ Driven Weather ...

    2020 doi

  13. [21]

    , Hoyer, S

    rasp_weatherbench_2024 APACrefauthors Rasp, S. , Hoyer, S. , Merose, A. , Langmore, I. , Battaglia, P. , Russel, T. Sha, F. APACrefauthors \ 2024 . WeatherBench 2: A benchmark for the next generation of data-driven global weather models. WeatherBench 2: A benchmark for the nex...

    2024 arXiv

  14. [22]

    \ Lean, H W

    roberts_scale-selective_2008 APACrefauthors Roberts, N M. \ Lean, H W. APACrefauthors \ 2008 . Scale- Selective Verification of Rainfall Accumulations from High - Resolution Forecasts of Convective Events Scale- Selective Verification of Rainfall Accumulations from High - Reso...

    2008 doi

  15. [23]

    , Richardson, D S

    rodwell_new_2010 APACrefauthors Rodwell, M J. , Richardson, D S. , Hewson, T D. \ Haiden, T. APACrefauthors \ 2010 . A new equitable score suitable for verifying precipitation in numerical weather prediction A new equitable score suitable for verifying precipitation in numeric...

    2010 doi

  16. [24]

    , Roy, S

    schmude_prithvi_2024 APACrefauthors Schmude, J. , Roy, S. , Trojak, W. , Jakubik, J. , Civitarese, D S. , Singh, S. Ramachandran, R. APACrefauthors \ 2024 . Prithvi WxC : Foundation Model for Weather and Climate . Prithvi WxC : Foundation Model for Weather and Climate . arXiv ...

    2024 doi arXiv

  17. [25]

    , Liu, S

    wang_orbit_2024 APACrefauthors Wang, X. , Liu, S. , Tsaris, A. , Choi, J Y. , Aji, A M. , Fan, M. Balaprakash, P. APACrefauthors \ 2024 . ORBIT : Oak ridge base foundation model for earth system predictability ORBIT : Oak ridge base foundation model for earth system predictabi...

    2024 arXiv

  18. [26]

    , Henn, B

    watt-meyer_ace2_2024 APACrefauthors Watt-Meyer, O. , Henn, B. , McGibbon, J. , Clark, S K. , Kwa, A. , Perkins, W A. Bretherton, C S. APACrefauthors \ 2024 . ACE2 : Accurately learning subseasonal to decadal atmospheric variability and forced responses. ACE2 : Accurately learn...

    2024 doi arXiv

  19. [27]

    , Zheng, X

    xu_artificial_2025 APACrefauthors Xu, P. , Zheng, X. , Gao, T. , Wang, Y. , Yin, J. , Zhang, J. Chen, X. APACrefauthors \ 2025 . An artificial intelligence-based limited area model for forecasting of surface meteorological variables An artificial intelligence-based limited are...

    2025 doi

  20. [28]

    , Zhou, Z

    zhao_weathergfm_2024 APACrefauthors Zhao, X. , Zhou, Z. , Zhang, W. , Liu, Y. , Chen, X. , Gong, J. Bai, L. APACrefauthors \ 2024 . WeatherGFM : Learning A Weather Generalist Foundation Model via In -context Learning . WeatherGFM : Learning A Weather Generalist Foundation Mode...

    2024 doi arXiv

Pith tools

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