Pith. sign in

REVIEW 2 major objections 48 references

Adding surface descriptors to a high-resolution data-driven weather model cuts 2 m temperature and 10 m wind errors, especially over cities and coasts, and lets glacier maps be updated without retraining.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-12 06:47 UTC pith:2EKJIZKV

load-bearing objection Solid ablation showing surface maps help near-surface AI forecasts, with a useful glacier-update demo; decoder-only training and MEPS target limit how far the operational claim can be pushed. the 2 major comments →

arxiv 2607.02824 v1 pith:2EKJIZKV submitted 2026-07-02 physics.ao-ph

Enhancing a high resolution data-driven weather prediction model with surface descriptors

{\AA}smund Bakketun , H{\aa}vard Homleid Haugen , Jostein Blyverket , Thomas Nils Nipen , Malte M\"uller This is my paper
classification physics.ao-ph
keywords data-driven weather predictionsurface descriptorstopographyland-atmosphere interactionsnear-surface variablesurban fractionglacier fractionhigh-resolution forecasting
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

High-resolution data-driven weather models still under-perform on near-surface variables such as 2-metre temperature and 10-metre wind, particularly over rare or heterogeneous surfaces. This paper shows that feeding a secondary decoder with static surface descriptors taken from the same land-surface scheme used by the training NWP system, plus simple topographic neighbourhood indices, measurably reduces those errors. Domain-wide mean absolute error falls by 1.9 % for temperature and 3.0 % for wind; over urban grid cells the temperature improvement reaches about 12 %. Removing glacier fraction at inference time produces a physically plausible warming, demonstrating that the descriptors are actually used and that evolving surface maps can be swapped in without an expensive full retrain. The practical claim is that surface information is not optional decoration but a necessary input if kilometre-scale data-driven forecasts of local weather are to be trusted.

Core claim

When a pretrained high-resolution data-driven model is given additional static surface descriptors (urban fraction, forest fraction, glacier fraction, soil texture, sub-grid orography, topographic position indices, etc.) through a secondary decoder, six-hour forecasts of 2 m temperature and 10 m wind improve relative to an otherwise identical baseline decoder that receives only the standard forcings of elevation, land–sea mask and astronomical quantities. The largest gains occur over under-sampled surfaces such as towns and coastlines, and the model responds to an artificial zeroing of glacier fraction by raising temperature over those cells, confirming that the descriptors are causally used

What carries the argument

A frozen encoder–processor backbone plus a newly trained secondary decoder that receives the extra surface descriptors only through the decoder’s query feature path, scored with almost-fair CRPS on 2 m temperature and 10 m wind over the Nordic high-resolution domain.

Load-bearing premise

That training only a secondary decoder while keeping the rest of the model frozen, and scoring against analyses produced by the same surface-scheme family, is enough to prove that surface descriptors improve real local physics rather than simply better matching the target model’s own surface-conditioned fields.

What would settle it

Retrain the full encoder–processor–decoder stack with and without the surface descriptors and evaluate both models against independent station observations over cities, glaciers and complex terrain; if the MAE reductions disappear or reverse, the central claim fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Operational centres can update glacier, urban and forest maps as they change without retraining the expensive backbone model.
  • Near-surface verification scores over towns and coastlines should rise once urban fraction and related descriptors are supplied as standard inputs.
  • Topographic neighbourhood indices give a modest but measurable gain specifically over mountain grid cells.
  • Static surface descriptors become a natural companion to any future inclusion of dynamic land variables such as snow or soil moisture.

Where Pith is reading between the lines

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

  • Because the largest gains appear on under-sampled surfaces, observation-only training pipelines that lack dense station coverage over cities may still need high-resolution surface maps to generalise correctly.
  • The same decoder-only experiment could be used as a cheap screen for which dynamic land variables (snow cover, soil moisture) are worth promoting to prognostic status in the full model.
  • If surface descriptors can be swapped at inference without retraining, climate-change scenarios that alter land cover become inexpensive sensitivity experiments for data-driven forecast systems.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 0 minor

Summary. The paper investigates whether adding static surface descriptors (SURFEX physiography fields and topographic neighbourhood indices) improves high-resolution forecasts of 2 m temperature and 10 m wind in the stretched-grid data-driven model Bris. To avoid full retraining, the authors freeze a pretrained encoder/processor and train only a secondary decoder that receives the extra forcings via the decoder query path. Four decoder configurations (BASELINE, SFX, TOPO, COMBINED) are trained under a fixed schedule and verified for one year against MEPS analysis. Domain-wide 6 h MAE falls by 1.9 % (T2m) and 3.0 % (wind); larger gains appear over town (~12 % T2m), coast and glacier. A glacier-zero inference experiment produces a physically plausible daytime temperature rise without retraining, supporting the claim that surface maps can be updated at inference time.

Significance. The work addresses a practical gap for kilometre-scale data-driven NWP: under-sampled surfaces (urban, glacier, forest) and the high cost of full retraining. The controlled decoder-only design, surface-stratified verification, and glacier ablation are clean and reproducible within the Anemoi framework. If the gains survive independent observation-based verification and full end-to-end training, the operational implication—that evolving surface maps can be swapped without re-training—would be valuable. The study therefore supplies useful evidence and a low-cost experimental template even if the absolute skill numbers are partly MEPS-matching.

major comments (2)
  1. Methods §2.1 and evaluation §3: all skill is measured against MEPS analysis, which itself is produced with SURFEX tiles and the same physiography family supplied as forcings (Tables 3–4). The reported MAE reductions (especially the ~12 % urban T2m gain) therefore partly quantify how faithfully the decoder remaps a frozen latent state onto MEPS’s own surface-conditioned fields rather than independent local physics. At least one verification against independent station or satellite observations over town/glacier is needed to support the claim of improved representation of real near-surface conditions.
  2. §2.1 and Discussion §4: the secondary decoder receives extra static forcings only through the query path; encoder and processor remain frozen. The experiment therefore cannot demonstrate that surface descriptors improve the full model’s latent dynamics or that the same maps would remain useful after end-to-end retraining. The operational claim that “input datasets can be updated without the need to retrain the model” should be explicitly scoped to decoder-only heads, or a limited full-model ablation should be added.

Circularity Check

0 steps flagged

No derivation-by-construction circularity; empirical decoder experiments against MEPS are self-contained (minor self-citation of base Bris architecture only).

full rationale

This is an empirical machine-learning paper, not a first-principles derivation. The load-bearing claims are measured MAE reductions (1.9 % / 3.0 % domain-wide; ~12 % T2m over town) obtained by training secondary decoders that receive extra static SURFEX-style and topographic forcings, then scoring 6 h forecasts against held-out MEPS analysis (Methods 2.1–2.3, Tables 2–4, Results 3.2–3.3). That is ordinary supervised learning with an ablation; the forecast is not algebraically forced by the fitted weights or by the input maps. The glacier-zero intervention (SFX G0, §3.5, Figs. 9–10) is a sensitivity test that demonstrates a physically plausible temperature rise when an input field is altered at inference; it does not redefine the skill metric. Self-citation of Nordhagen et al. (2025) supplies only the frozen base encoder/processor architecture and is not used to import a uniqueness theorem or to forbid alternatives. Evaluation against MEPS (which itself uses the same physiography family) raises a legitimate external-validity question about whether the gains reflect real local physics versus better MEPS-matching, but that is a correctness/benchmark concern, not circularity of the derivation chain under the stated criteria. No self-definitional equations, no fitted parameter renamed as an independent prediction, and no ansatz smuggled via citation appear. Score 1 only for the non-load-bearing architectural self-citation; the central experimental claims stand independently.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 1 invented entities

The claim rests on standard ML-weather training practice plus domain assumptions that MEPS analysis is a valid verification target and that decoder-only fine-tuning isolates surface-descriptor value. Free parameters are ordinary training and design choices (loss weights, CRPS ε, neighbourhood kernel, learning schedule). No new physical entities are postulated; surface descriptors are inherited from SURFEX/MEPS and topo filters.

free parameters (4)
  • almost-fair CRPS ε and ensemble size M = M=2, ε=0.025
    M=2 and ε=0.025 set the probabilistic loss balance; chosen for consistency with prior Bris/AIFS literature, not derived here.
  • variable loss weights (T vs wind components) = 1.0 (T), 0.1 (u,v)
    Temperature weight 1.0 and wind components 0.1 are hyperparameters that shape which errors the decoder optimizes.
  • topography neighbourhood kernel diameter = 12.5 km
    12.5 km (5-grid-point) neighbourhood for topo descriptors is a design choice that defines the local context features.
  • decoder training schedule (LR, warmup, epochs/steps, batch) = LR 1e-3; 70 epochs / 15000 steps; batch 16
    AdamW, 1e-3 LR with warmup+cosine, 70 epochs/15000 steps, batch 16 are fitted/chosen training settings that determine the reported skill.
axioms (4)
  • domain assumption MEPS operational analysis is an adequate ground truth for scoring 2 m temperature and 10 m wind skill of the data-driven decoder.
    All MAE/bias results in §3 are against MEPS; observational verification is not the primary score.
  • ad hoc to paper Freezing the pretrained encoder and processor while training only Decoder B still measures the importance of surface descriptors for high-resolution near-surface prediction.
    Stated cost-saving design in §2.1; authors note full-chain inclusion is likely better but do not test it.
  • domain assumption Static SURFEX physiography fields and derived topo indices are valid forcings that can be updated at inference without invalidating the trained decoder.
    Underpins the glacier zero-out interpretation and operational ‘no retrain’ claim in §3.5 and §4.
  • ad hoc to paper Single-member year-long averages cancel stochastic processor noise sufficiently for MAE comparisons.
    Explicit cost assumption in §2.3 for the main verification year.
invented entities (1)
  • Secondary decoder (Decoder B) attached to frozen Bris no independent evidence
    purpose: Allow low-cost ablation of extra surface forcings for Nordic 2 m T and 10 m wind without full retrain.
    Architectural experiment device, not a new physical object; independent evidence is only the reported skill deltas on MEPS.

pith-pipeline@v1.1.0-grok45 · 19369 in / 3485 out tokens · 34363 ms · 2026-07-12T06:47:49.572089+00:00 · methodology

0 comments
read the original abstract

We study the importance of surface characteristics when forecasting near-surface variables with a data-driven weather prediction model. To target the challenge of predicting small-scale weather conditions at high resolution, we introduce a range of surface descriptors in the training of a state-of-the-art data-driven model. The input data includes surface descriptors inherited from the numerical weather prediction model used to produce the training dataset and topographic neighbourhood indices. We found that errors of 2-metre temperature and 10-metre wind speed forecasts were reduced by 1.9% and 3.0% respectively compared to a baseline model over the model domain. Over certain surfaces, the improvements were significantly larger. For example, we found a 12% reduction of temperature mean absolute errors over urban areas when the urban fraction was included in the model input. Furthermore, we investigated how the model responded to removal of glaciers, resulting in an increase of temperature. This indicates that 1) the model produce a physically reasonable response and 2) input datasets can be updated without the need to retrain the model. The latter suggests a great benefit for operational systems as training is expensive compared to running these models. This study highlights the importance of including surface conditions in the prediction of near-surface variables.

Figures

Figures reproduced from arXiv: 2607.02824 by {\AA}smund Bakketun, H{\aa}vard Homleid Haugen, Jostein Blyverket, Malte M\"uller, Thomas Nils Nipen.

Figure 1
Figure 1. Figure 1: Model component diagram with examples of [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: 2-metre temperature mean difference (a) and [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 5
Figure 5. Figure 5: MAE against MEPS for (a) 2-metre temperature [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Same as Fig- 3 but for 10-metre wind speed [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 9
Figure 9. Figure 9: Map of original Glacier field (a) and 2-metre [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Difference between SFXG0 and SFX 2-metre [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

48 extracted references · 25 canonical work pages

  1. [1]

    Becker and G

    E. Becker and G. Schmitz , title =. J. Atmos. Sci. , year =

  2. [2]

    Kuji and T

    M. Kuji and T. Nakajima , YEAR = 2002, TITLE =. 11th Conf. on Cloud Physics , ADDRESS =

    2002

  3. [3]

    Meixner and L

    T. Meixner and L. A. Bastidas and H. V. Gupta and R. C. Bales , title =. Water Resour. Res. , volume =. doi:10.1029/2000WR000112 , eid =

    work page doi:10.1029/2000wr000112

  4. [4]

    Knutti and M

    R. Knutti and M. R. Allen and P. Friedlingstein and J. M. Gregory and G. C. Hegerl and G. A. Meeh and M. Meinshausen and J. M. Murphy and G.-K. Plattne and S. C. B. Raper and T. F. Stocker and P. A. Stott and H. Teng and and T. M. L. Wigley , title=. J. Climate , year =

  5. [5]

    Emery, W. J. and Thomas, A. C. and Collins, M. J. and Crawford, W. R. and Mackas, D. L. , title =. J. Geophys. Res. , volume =. doi:10.1029/JC091iC11p12865 , pages =

    work page doi:10.1029/jc091ic11p12865

  6. [6]

    International Journal of Climatology , author =

    Fine scale surface climate in complex terrain using machine learning , volume =. International Journal of Climatology , author =. 2021 , keywords =. doi:10.1002/joc.6617 , abstract =

    work page doi:10.1002/joc.6617 2021

  7. [7]

    Forecasting

    Keisler, Ryan , month = feb, year =. Forecasting. doi:10.48550/arXiv.2202.07575 , abstract =

    Pith/arXiv arXiv doi:10.48550/arxiv.2202.07575

  8. [8]

    doi:10.48550/arXiv.2212.12794 , abstract =

    Lam, Remi and Sanchez-Gonzalez, Alvaro and Willson, Matthew and Wirnsberger, Peter and Fortunato, Meire and Alet, Ferran and Ravuri, Suman and Ewalds, Timo and Eaton-Rosen, Zach and Hu, Weihua and Merose, Alexander and Hoyer, Stephan and Holland, George and Vinyals, Oriol and Stott, Jacklynn and Pritzel, Alexander and Mohamed, Shakir and Battaglia, Peter ...

    Pith/arXiv arXiv doi:10.48550/arxiv.2212.12794

  9. [9]

    Remote Sensing of Environment , author =

    A machine learning model for estimating the temperature of small rivers using satellite-based spatial data , volume =. Remote Sensing of Environment , author =. 2024 , keywords =. doi:10.1016/j.rse.2024.114271 , abstract =

    work page doi:10.1016/j.rse.2024.114271 2024

  10. [10]

    2025 , note =

    Communications Earth & Environment , author =. 2025 , note =. doi:10.1038/s43247-025-02502-y , abstract =

    doi:10.1038/s43247-025-02502-y 2025

  11. [11]

    Regional data-driven weather modeling with a global stretched-grid , volume =

    Nipen, Thomas Nils and Haugen, Håvard Homleid and Ingstad, Magnus Sikora and Nordhagen, Even Marius and Salihi, Aram Farhad Shafiq and Tedesco, Paulina and Seierstad, Ivar Ambjørn and Kristiansen, Jørn and Lang, Simon and Alexe, Mihai and Dramsch, Jesper and Raoult, Baudouin and Mertes, Gert and Chantry, Matthew , month = nov, year =. Regional data-driven...

    doi:10.1175/aies-d-25-0001.1

  12. [12]

    A deep learning framework for predicting the effect of surface topography on thermal contact resistance , volume =

    Zhou, Man and He, Zhuoyan and Guo, Peiyao and Zhang, Ping and Tao, Lin and Chen, Mengjun , month = oct, year =. A deep learning framework for predicting the effect of surface topography on thermal contact resistance , volume =. Communications Engineering , publisher =. doi:10.1038/s44172-025-00508-0 , abstract =

    work page doi:10.1038/s44172-025-00508-0

  13. [13]

    arXiv preprint arXiv:2009.03509 , year=

    Masked label prediction: Unified message passing model for semi-supervised classification , author=. arXiv preprint arXiv:2009.03509 , year=

    Pith/arXiv arXiv 2009

  14. [14]

    and Yang, Jing and Zadra, Ayrton , month = aug, year =

    Husain, Syed Zahid and Separovic, Leo and Caron, Jean-François and Aider, Rabah and Buehner, Mark and Chamberland, Stéphane and Lapalme, Ervig and McTaggart-Cowan, Ron and Subich, Christopher and Vaillancourt, Paul A. and Yang, Jing and Zadra, Ayrton , month = aug, year =. Leveraging. Weather and Forecasting , publisher =. doi:10.1175/WAF-D-24-0139.1 , abstract =

    doi:10.1175/waf-d-24-0139.1

  15. [15]

    Machine Learning with Applications , author =

    Data-driven regional weather forecasting:. Machine Learning with Applications , author =. 2026 , keywords =. doi:10.1016/j.mlwa.2026.100887 , abstract =

    work page doi:10.1016/j.mlwa.2026.100887 2026

  16. [16]

    A regional high resolution

    Baño-Medina, Jorge and Sengupta, Agniv and Steinhoff, Daniel and Mulrooney, Patrick and Nipen, Thomas and Santa-Cruz, Mario and Nie, Yanbo and Delle Monache, Luca , month = dec, year =. A regional high resolution. npj Climate and Atmospheric Science , publisher =. doi:10.1038/s41612-025-01265-9 , abstract =

    work page doi:10.1038/s41612-025-01265-9

  17. [17]

    , month = jun, year =

    Balsamo, Gianpaolo and Beljaars, Anton and Scipal, Klaus and Viterbo, Pedro and Hurk, Bart van den and Hirschi, Martin and Betts, Alan K. , month = jun, year =. A. Journal of Hydrometeorology , publisher =. doi:10.1175/2008JHM1068.1 , abstract =

    work page doi:10.1175/2008jhm1068.1

  18. [18]

    Improvements in the land surface configuration to better simulate seasonal snow cover in the

    Monteiro, Diego and Caillaud, Cécile and Lafaysse, Matthieu and Napoly, Adrien and Fructus, Mathieu and Alias, Antoinette and Morin, Samuel , month = nov, year =. Improvements in the land surface configuration to better simulate seasonal snow cover in the. Geoscientific Model Development , publisher =. doi:10.5194/gmd-17-7645-2024 , abstract =

    work page doi:10.5194/gmd-17-7645-2024 2024

  19. [19]

    Quarterly Journal of the Royal Meteorological Society , author =

    The energetic basis of the urban heat island , volume =. Quarterly Journal of the Royal Meteorological Society , author =. 1982 , pages =. doi:10.1002/qj.49710845502 , language =

    work page doi:10.1002/qj.49710845502 1982

  20. [20]

    doi:10.48550/arXiv.2202.11214 , abstract =

    Pathak, Jaideep and Subramanian, Shashank and Harrington, Peter and Raja, Sanjeev and Chattopadhyay, Ashesh and Mardani, Morteza and Kurth, Thorsten and Hall, David and Li, Zongyi and Azizzadenesheli, Kamyar and Hassanzadeh, Pedram and Kashinath, Karthik and Anandkumar, Animashree , month = feb, year =. doi:10.48550/arXiv.2202.11214 , abstract =

    Pith/arXiv arXiv doi:10.48550/arxiv.2202.11214

  21. [21]

    and Noer, Gunnar and Korosov, Anton , month = dec, year =

    Müller, Malte and Batrak, Yurii and Kristiansen, Jørn and Køltzow, Morten A. and Noer, Gunnar and Korosov, Anton , month = dec, year =. Characteristics of a. Monthly Weather Review , publisher =. doi:10.1175/MWR-D-17-0194.1 , language =

    work page doi:10.1175/mwr-d-17-0194.1

  22. [22]

    Weather and Forecasting , publisher =

    Frogner, Inger-Lise and Andrae, Ulf and Bojarova, Jelena and Callado, Alfons and Escribà, Pau and Feddersen, Henrik and Hally, Alan and Kauhanen, Janne and Randriamampianina, Roger and Singleton, Andrew and Smet, Geert and van der Veen, Sibbo and Vignes, Ole , month = dec, year =. Weather and Forecasting , publisher =. doi:10.1175/WAF-D-19-0030.1 , language =

    doi:10.1175/waf-d-19-0030.1

  23. [23]

    Bengtsson, Lisa and Andrae, Ulf and Aspelien, Trygve and Batrak, Yurii and Calvo, Javier and de Rooy, Wim and Gleeson, Emily and Hansen-Sass, Bent and Homleid, Mariken and Hortal, Mariano and Ivarsson, Karl-Ivar and Lenderink, Geert and Niemelä, Sami and Nielsen, Kristian Pagh and Onvlee, Jeanette and Rontu, Laura and Samuelsson, Patrick and Muñoz, Daniel...

    work page doi:10.1175/mwr-d-16-0417.1

  24. [24]

    and Moigne, P

    Masson, V. and Moigne, P. Le and Martin, E. and Faroux, S. and Alias, A. and Alkama, R. and Belamari, S. and Barbu, A. and Boone, A. and Bouyssel, F. and Brousseau, P. and Brun, E. and Calvet, J.-C. and Carrer, D. and Decharme, B. and Delire, C. and Donier, S. and Essaouini, K. and Gibelin, A.-L. and Giordani, H. and Habets, F. and Jidane, M. and Kerdraon...

    work page doi:10.5194/gmd-6-929-2013 2013

  25. [25]

    Masson, Valéry and Champeaux, Jean-Louis and Chauvin, Fabrice and Meriguet, Christelle and Lacaze, Roselyne , month = may, year =. A. Journal of Climate , publisher =. doi:10.1175/1520-0442(2003)16<1261:AGDOLS>2.0.CO;2 , abstract =

    doi:10.1175/1520-0442(2003)16 2003

  26. [26]

    and Kaptué Tchuenté, A

    Faroux, S. and Kaptué Tchuenté, A. T. and Roujean, J.-L. and Masson, V. and Martin, E. and Le Moigne, P. , month = apr, year =. Geoscientific Model Development , publisher =. doi:10.5194/gmd-6-563-2013 , abstract =

    work page doi:10.5194/gmd-6-563-2013 2013

  27. [27]

    Global and Planetary Change , author =

    The. Global and Planetary Change , author =. 1996 , pages =. doi:10.1016/0921-8181(95)00043-7 , abstract =

    work page doi:10.1016/0921-8181(95)00043-7 1996

  28. [28]

    The interactions between soil–biosphere–atmosphere land surface model with a multi-energy balance (

    Boone, Aaron and Samuelsson, Patrick and Gollvik, Stefan and Napoly, Adrien and Jarlan, Lionel and Brun, Eric and Decharme, Bertrand , month = feb, year =. The interactions between soil–biosphere–atmosphere land surface model with a multi-energy balance (. Geoscientific Model Development , publisher =. doi:10.5194/gmd-10-843-2017 , abstract =

    work page doi:10.5194/gmd-10-843-2017 2017

  29. [29]

    2026 , note =

    Meteo-Swiss , url =. 2026 , note =

    2026

  30. [30]

    How inaccurate offshore wind observations impact the quality of operational numerical weather prediction , volume =

    Køltzow, Morten and Øiestad, Merete and Azad, Roohollah and Furevik, Birgitte Rugaard and Olsen, Anne-Mette and Remes, Teresa , month = oct, year =. How inaccurate offshore wind observations impact the quality of operational numerical weather prediction , volume =. Quarterly Journal of the Royal Meteorological Society , publisher =. doi:10.1002/qj.4872 , ...

    work page doi:10.1002/qj.4872

  31. [31]

    A deep learning-based land-atmosphere coupled model for heatwave prediction , volume =

    Cho, Dongjin and Ham, Yoo-Geun and Jeong, Suyeon and Kang, Seon-Yu , month = feb, year =. A deep learning-based land-atmosphere coupled model for heatwave prediction , volume =. npj Climate and Atmospheric Science , publisher =. doi:10.1038/s41612-025-01311-6 , abstract =

    work page doi:10.1038/s41612-025-01311-6

  32. [32]

    Impacts of snow and organic soils parameterization on northern

    Decharme, Bertrand and Brun, Eric and Boone, Aaron and Delire, Christine and Le Moigne, Patrick and Morin, Samuel , month = apr, year =. Impacts of snow and organic soils parameterization on northern. The Cryosphere , publisher =. doi:https://doi.org/10.5194/tc-10-853-2016 , abstract =

    doi:10.5194/tc-10-853-2016 2016

  33. [33]

    Journal of Geophysical Research: Atmospheres , author =

    Local evaluation of the. Journal of Geophysical Research: Atmospheres , author =. 2011 , keywords =. doi:https://doi.org/10.1029/2011JD016002 , abstract =

    work page doi:10.1029/2011jd016002 2011

  34. [34]

    Atmospheric Research , author =

    Impact of snow cover on the surface energy balance and its contribution to the extreme cold wave in. Atmospheric Research , author =. 2025 , keywords =. doi:10.1016/j.atmosres.2025.108015 , abstract =

    work page doi:10.1016/j.atmosres.2025.108015 2025

  35. [35]

    and Andersson, Tom R

    Allen, Anna and Markou, Stratis and Tebbutt, Will and Requeima, James and Bruinsma, Wessel P. and Andersson, Tom R. and Herzog, Michael and Lane, Nicholas D. and Chantry, Matthew and Hosking, J. Scott and Turner, Richard E. , month = may, year =. End-to-end data-driven weather prediction , volume =. Nature , publisher =. doi:10.1038/s41586-025-08897-0 , a...

    doi:10.1038/s41586-025-08897-0

  36. [36]

    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

  37. [37]

    arXiv preprint arXiv:2309.17370 , year=

    Graph-based neural weather prediction for limited area modeling , author=. arXiv preprint arXiv:2309.17370 , year=

    Pith/arXiv arXiv

  38. [38]

    ecmwf/anemoi-core , copyright =

    Anemoi-Contributors , month = apr, year =. ecmwf/anemoi-core , copyright =

  39. [39]

    Quarterly Journal of the Royal Meteorological Society , author =

    The. Quarterly Journal of the Royal Meteorological Society , author =. 2020 , keywords =. doi:10.1002/qj.3803 , abstract =

    doi:10.1002/qj.3803 2020

  40. [40]

    Earth-Science Reviews , author =

    Investigating soil moisture–climate interactions in a changing climate:. Earth-Science Reviews , author =. 2010 , keywords =. doi:10.1016/j.earscirev.2010.02.004 , abstract =

    work page doi:10.1016/j.earscirev.2010.02.004 2010

  41. [41]

    Boundary-Layer Meteorology , author =

    A parametric model of vertical eddy fluxes in the atmosphere , volume =. Boundary-Layer Meteorology , author =. 1979 , keywords =. doi:10.1007/BF00117978 , abstract =

    work page doi:10.1007/bf00117978 1979

  42. [42]

    Smith, B, Ronald , month = jan, year =. The. Advances in. doi:10.1016/S0065-2687(08)60262-9 , abstract =

    work page doi:10.1016/s0065-2687(08)60262-9

  43. [43]

    Bouallègue, Zied Ben and Clare, Mariana C. A. and Magnusson, Linus and Gascón, Estibaliz and Maier-Gerber, Michael and Janoušek, Martin and Rodwell, Mark and Pinault, Florian and Dramsch, Jesper S. and Lang, Simon T. K. and Raoult, Baudouin and Rabier, Florence and Chevallier, Matthieu and Sandu, Irina and Dueben, Peter and Chantry, Matthew and Pappenberg...

    doi:10.1175/bams-d-23-0162.1

  44. [44]

    Journal of Geophysical Research: Machine Learning and Computation , author =

    Comparison of. Journal of Geophysical Research: Machine Learning and Computation , author =. 2025 , pages =. doi:10.1029/2024JH000481 , abstract =

    work page doi:10.1029/2024jh000481 2025

  45. [45]

    Automated

    Malinowski, Radek and Lewiński, Stanisław and Rybicki, Marcin and Gromny, Ewa and Jenerowicz, Małgorzata and Krupiński, Michał and Nowakowski, Artur and Wojtkowski, Cezary and Krupiński, Marcin and Krätzschmar, Elke and Schauer, Peter , month = jan, year =. Automated. Remote Sensing , publisher =. doi:10.3390/rs12213523 , abstract =

    work page doi:10.3390/rs12213523

  46. [46]

    and Mathis, Mark and Brumby, Steven P

    Karra, Krishna and Kontgis, Caitlin and Statman-Weil, Zoe and Mazzariello, Joseph C. and Mathis, Mark and Brumby, Steven P. , month = jul, year =. Global land use / land cover with. 2021. doi:10.1109/IGARSS47720.2021.9553499 , abstract =

    doi:10.1109/igarss47720.2021.9553499 2021

  47. [47]

    Nordhagen, Even Marius and Haugen, Håvard Homleid and Salihi, Aram Farhad Shafiq and Ingstad, Magnus Sikora and Nipen, Thomas Nils and Seierstad, Ivar Ambjørn and Frogner, Inger-Lise and Clare, Mariana and Lang, Simon and Chantry, Matthew and Dueben, Peter and Kristiansen, Jørn , month = nov, year =. High-. doi:10.48550/arXiv.2511.23043 , abstract =

    doi:10.48550/arxiv.2511.23043

  48. [48]

    and Planton, S

    Noilhan, J. and Planton, S. , month = mar, year =. A. Monthly Weather Review , publisher =. doi:10.1175/1520-0493(1989)117<0536:ASPOLS>2.0.CO;2 , abstract =

    doi:10.1175/1520-0493(1989)117 1989