REVIEW 3 major objections 5 minor 41 references
A frozen global weather foundation model can drive accurate regional forecasts at ~100 imes finer grid resolution using only lightweight latent heads.
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 →
T0 review · grok-4.5
2026-07-12 03:34 UTC pith:NRG5DKOJ
load-bearing objection Solid systems result: frozen Aurora + multi-scale latent heads give LBC-free ~0.025° CONUS downscaling that often beats WRF-ARW and image SR at tiny cost; the NWP-target/bias caveat is real but does not erase the empirical contribution. the 3 major comments →
From Global to Local: Efficient Regional Weather Downscaling with Global Weather Foundation Model
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
A frozen 0.25° Aurora backbone, augmented only with lightweight multi-scale prediction heads that operate directly in its latent space, supports regional adaptation at resolutions corresponding to a two-order-of-magnitude increase in grid-cell resolution and improves accuracy versus NWP (WRF-ARW) on most metrics at a fraction of the computational cost, while outperforming standard image-based super-resolution downscaling.
What carries the argument
Latent-space multi-scale decoder heads (MLP-S-L and parallel/sequential compositions) that map Aurora’s frozen 1024-dimensional patch embeddings to high-resolution surface fields; the global coupling stays inside the frozen backbone while decoding remains patch-local and modular.
Load-bearing premise
The frozen global embeddings already contain enough multi-scale information that simple heads trained only on regional numerical products can recover true mesoscale structure without fine-tuning the backbone or enforcing physical constraints.
What would settle it
Train the same heads on a new geographic domain or against independent high-resolution reanalysis that was never used in training; if RMSE versus stations and energy spectra degrade sharply relative to the CONUS/HRRR results, the claim that the frozen latent space generalizes fails.
If this is right
- Operational regional forecasting can drop NWP-based local boundary conditions and still remain consistent with global dynamics.
- A single frozen global backbone can serve many regions via cheap, independently trained decoder heads.
- Inference cost drops from hours of HPC simulation to roughly one minute per year on a single GPU.
- Latent-space adaptation outperforms pure image super-resolution for atmospheric fields, suggesting foundation-model embeddings are the right intermediate representation for downscaling.
- Rollout error growth over 48 h is substantially lower than for the corresponding NWP forecasts on the reported variables.
Where Pith is reading between the lines
- If the same heads work for precipitation (omitted in the original backbone), the method could become a full surface-weather regional engine without any mesoscale NWP.
- The multi-region training result (CONUS + Denmark) hints that a single set of heads might cover continental-scale domains, reducing the need for region-specific models.
- Bias inherited from the regional training targets (WTK-US vs HRRR) remains visible; future work that jointly debiases targets and heads could close the remaining temperature gap versus NWP on gridded evaluation.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes a regional weather downscaling framework that freezes the Aurora 0.25° foundation-model backbone and trains only lightweight multi-scale decoder heads (MLP-S-L, parallel multi-resolution compositions, optional orography encoders) in its latent space to map ERA5-derived embeddings to ~0.025° surface fields (T2, u10, v10, surface pressure). Training targets are regional NWP products (WTK-US, HRRR); evaluation uses the HRRR grid, HadISD stations, energy spectra, multi-region training, and 48 h rollouts against WRF-ARW, Aurora 0.25°, StormCast, and image-based super-resolution baselines. The central claim is that this latent-space adaptation recovers fine-scale structure, improves accuracy versus WRF-ARW on most metrics at a fraction of the cost, and outperforms standard super-resolution approaches without backbone retraining or NWP local boundary conditions.
Significance. If the result holds, the work offers a practical and computationally attractive route to LBC-free regional forecasting by reusing a frozen global foundation model rather than training or fine-tuning full regional models. Strengths include systematic ablations of embedding-based versus patch-based heads (Tables 1–2), external baselines including operational WRF-ARW and StormCast (Table 3), independent HadISD station evaluation, energy-spectra checks (Appendix F), multi-region training (Appendix H), and rollout stability (Figure 3). These elements make the contribution concrete and falsifiable for operational downscaling and foundation-model adaptation in weather ML.
major comments (3)
- The headline claim of improved accuracy versus NWP (Abstract; §5; Table 3) is load-bearing on WTK-US/HRRR as high-resolution targets. Appendix D documents systematic temperature biases of WTK-US (−0.63 °C) and HRRR (−0.29 °C) relative to ERA5, and the authors themselves attribute the T2 underperformance versus WRF-ARW on the HRRR grid to inherited WTK-US bias. Because the primary gridded targets and the WRF-ARW baseline share modeling lineage, it remains unclear how much of the reported skill is bias/regridding matching rather than recovery of physical mesoscale structure. A clearer separation—e.g., bias-corrected targets, evaluation against independent high-resolution analyses, or explicit residual-bias diagnostics—is needed before the NWP-beating claim can be taken at face value.
- Appendix B reports large round-trip regridding error for surface pressure (RMSE 2.34 ± 40.4 hPa), while Tables 1 and 3 present some of the largest claimed gains on P. Given the documented HRRR Great Lakes elevation artifact (Appendix C) and the exclusion of Dec 2020–Jul 2022 for pressure, the pressure improvements are especially sensitive to grid alignment and elevation handling. The manuscript should quantify how much of the P skill survives under alternative regridding schemes or on native LCC grids, and should temper claims that rest primarily on pressure RMSE.
- Method §4 and Limitations §6 state that heads operate without explicit physical constraints; Appendix F shows spectra consistent with Aurora only up to its effective resolution, after which tails can reflect NWP-specific fine-scale artifacts. Combined with nearest-grid-point HadISD matching and sparse station coverage (§3, Appendix G), the evidence that the heads recover true mesoscale dynamics—rather than the statistical signature of the regional NWP products—is incomplete. Additional diagnostics (e.g., case studies of known mesoscale features, cross-validation on held-out regions/years not used as targets, or comparison to independent observational analyses) would strengthen the central physical claim.
minor comments (5)
- Figure 1 caption and §4 notation for decoder heads (MLP-S-L, |, +, ↑40 NN) are dense; a short explicit glossary or table of all evaluated compositions would improve readability.
- Table 1 marks Aurora pressure as unavailable with a footnote; the same constraint should be stated once in the experimental-setup paragraph to avoid repeated asterisks.
- Appendix H multi-region results (CONUS + DANRA) are encouraging but use different native resolutions and temporal windows; a one-sentence note on how regridding and lead-time alignment were handled would help.
- Typos and style: “neareast neighbour”, “out work”, “mlp-based”, and mixed en-dashes/hyphens appear in §4–5; a light copy-edit pass is warranted.
- Figure 3 rollout horizon is incomplete for WRF-ARW pressure (only to 36 h); the caption should state this limitation explicitly rather than only in the main text.
Circularity Check
No significant circularity: skill is measured against external NWP products and independent stations, not defined by construction from the same quantity predicted.
full rationale
The paper trains lightweight multi-scale decoder heads on frozen Aurora latent embeddings (from ERA5) against regional NWP targets (WTK-US, HRRR) and evaluates RMSE against those grids, WRF-ARW forecasts, and independent HadISD stations. Nothing in the method defines the predicted fields or the claimed skill via a fitted parameter that is then re-reported as a prediction, nor via a uniqueness theorem or self-citation chain that forces the result. Aurora is an external foundation model; the heads are ordinary supervised adapters. The usual literature risk that training targets share lineage with the WRF-ARW baseline (and that regridding/bias can dominate) is a correctness/evaluation concern, not circularity by construction. Score 1 only for the mild, non-load-bearing fact that the primary gridded targets and one baseline are both NWP products; the central claim remains independently testable on HadISD and against non-NWP baselines (super-resolution, Aurora base, StormCast).
Axiom & Free-Parameter Ledger
free parameters (4)
- Decoder architecture (S, L, parallel/sequential head composition)
- AdamW learning rate schedule
- Orography encoding variant (raw H vs learned LH)
- Training data mix and temporal windows (WTK-US + HRRR; pressure 2017–2019 train / 2020 eval with Dec 2020–Jul 2022 exclu
axioms (5)
- domain assumption Aurora’s frozen backbone embeddings encode multi-scale atmospheric structure sufficient for regional refinement without backbone updates.
- domain assumption Regional NWP products (WTK-US, HRRR) after physically aware regridding are valid high-resolution training targets for surface T, wind, and pressure.
- domain assumption Patch-local decoding of globally contextualized embeddings yields spatially uniform skill without LBC forcing.
- ad hoc to paper Nearest-neighbor/bicubic deterministic upsampling of coarser head outputs to 40×40 is an acceptable non-learned interface for multi-scale composition.
- domain assumption Standard supervised MAE training with independent per-variable heads is adequate; explicit conservation or PDE constraints are not required for the claimed skill.
invented entities (1)
-
Multi-scale latent decoder head family (MLP-S-L, MLPP-S-L, parallel | and sequential + compositions, LH/H orography encoders)
no independent evidence
Cite this review
Pith. "Pith review of From Global to Local: Efficient Regional Weather Downscaling with Global Weather Foundation Model." pith.science (2026). https://pith.science/paper/NRG5DKOJ
@misc{pith2026260703279,
author = {Pith},
title = {Pith review of: From Global to Local: Efficient Regional Weather Downscaling with Global Weather Foundation Model},
year = {2026},
howpublished = {\url{https://pith.science/paper/NRG5DKOJ}},
note = {Machine review of arXiv:2607.03279}
}
read the original abstract
Accurate regional weather prediction requires resolving fine-scale structure while remaining consistent with global dynamics. Traditional limited area models rely on computationally expensive simulations, while many learning-based approaches frame the problem as super-resolution, overlooking statistical and physical mismatches across scales. We propose a foundation-model-driven downscaling framework that learns regional refinements of global forecasts by augmenting a pretrained weather model backbone with lightweight, multi-scale prediction heads operating directly in its latent space. Despite being trained on substantially coarser inputs, the pretrained backbone supports regional adaptation at resolutions corresponding to a two-order-of-magnitude increase in grid-cell resolution, without the need for retraining. The proposed approach uses regional numerical simulations as training targets and is evaluated not only against gridded datasets but also against ground-based weather station observations, enabling analysis of systematic biases between global reanalysis, regional simulations, and in-situ weather station observations. Our experiments show improved accuracy in comparison to NWP on most of the metrics at the fraction of computational cost. Moreover, we observe that building on a latent space of globally pre-trained weather foundation model offers better downscaling capabilities than the standard image-based super-resolution approaches.
Figures
Reference graph
Works this paper leans on
-
[1]
by Lewis F
Weather prediction by numerical process. by Lewis F. Richardson. cambridge (university press), 1922. 4°. pp. xii + 236. 30s.net.Quarterly Journal of the Royal Meteorological Society, 48(203):282–284, 1922
1922
-
[2]
Maier and Graeme C
Holger R. Maier and Graeme C. Dandy. Neural networks for the prediction and forecasting of water resources variables: a review of modelling issues and applications.Environmental Modelling & Software, 15(1):101–124, 2000
2000
-
[3]
Papagiannaki, K
K. Papagiannaki, K. Lagouvardos, V . Kotroni, and A. Bezes. Flash flood occurrence and relation to the rainfall hazard in a highly urbanized area.Natural Hazards and Earth System Sciences, 15(8):1859–1871, 2015
2015
-
[4]
Olatunji, Eyitayo Adeoye, Tien-Chien Jen, and Daniel M
Kingsley Ukoba, Kehinde O. Olatunji, Eyitayo Adeoye, Tien-Chien Jen, and Daniel M. Madyira. Optimizing renewable energy systems through artificial intelligence: Review and future prospects.Energy & Environment, 35(7):3833–3879, 2024
2024
-
[5]
A description of the advanced research wrf version 3, 06 2008
William Skamarock, Joseph Klemp, Jimy Dudhia, David Gill, Dale Barker, Michael Duda, Xiang-Yu Huang, Wei Wang, and Jordan Powers. A description of the advanced research wrf version 3, 06 2008
2008
-
[6]
Skamarock, Joseph B
William C. Skamarock, Joseph B. Klemp, Jimy Dudhia, David O. Gill, Zhiquan Liu, Judith Berner, Wei Wang, John G. Powers, Michael G. Duda, Dale M. Barker, and Xiang-Yu Huang. A description of the advanced research wrf version 4. Technical Report NCAR/TN-556+STR, National Center for Atmospheric Research, 2019
2019
-
[7]
Zaviša I. Janjic. A nonhydrostatic mesoscale model. part i: Dynamics.Meteorology and Atmospheric Physics, 82(1–4):147–169, 2003
2003
-
[8]
Bruinsma, Tom R
Anna Vaughan, Stratis Markou, Will Tebbutt, James Requeima, Wessel P. Bruinsma, Tom R. Andersson, Michael Herzog, Nicholas D. Lane, Matthew Chantry, J. Scott Hosking, and Richard E. Turner. Aardvark weather: end-to-end data-driven weather forecasting, 2024
2024
-
[9]
Learning skillful medium-range global weather forecasting
Remi Lam, Alvaro Sanchez-Gonzalez, Matthew Willson, Peter Wirnsberger, Meire Fortunato, Ferran Alet, Suman Ravuri, Timo Ewalds, Zach Eaton-Rosen, Weihua Hu, Alexander Merose, Stephan Hoyer, George Holland, Oriol Vinyals, Jacklynn Stott, Alexander Pritzel, Shakir Mohamed, and Peter Battaglia. Learning skillful medium-range global weather forecasting. Scien...
2023
-
[10]
A foundation model for the earth system
Cristian Bodnar, Wessel P Bruinsma, Ana Lucic, Megan Stanley, Anna Allen, Johannes Brand- stetter, Patrick Garvan, Maik Riechert, Jonathan A Weyn, Haiyu Dong, et al. A foundation model for the earth system. https://github.com/microsoft/aurora/tree/main/aurora, 2025. Released under the MIT License
2025
-
[11]
Jaideep Pathak, Shashank Subramanian, Peter Harrington, Sanjeev Raja, Ashesh Chattopadhyay, Morteza Mardani, Thorsten Kurth, David Hall, Zongyi Li, Kamyar Azizzadenesheli, et al. Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators.arXiv preprint arXiv:2202.11214, 2022
Pith/arXiv arXiv 2022
-
[12]
Hans Hersbach, Bill Bell, Paul Berrisford, Shoji Hirahara, András Horányi, Joaquín Muñoz- Sabater, Julien Nicolas, Caroline Peubey, Raluca Radu, Dinand Schepers, Adrian Simmons, Claudia Soci, Saleh Abdalla, Xavier Abellan, Gianpaolo Balsamo, Peter Bechtold, Giovanna Biavati, Jean-Raymond Bidlot, Massimo Bonavita, Giovanna De Chiara, Per Dahlgren, Dick Dee...
-
[13]
Accessed via the Copernicus Climate Data Store, Released under Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
-
[14]
Simon Lang, Mihai Alexe, Matthew Chantry, Jesper Dramsch, Florian Pinault, Baudouin Raoult, Mariana C. A. Clare, Christian Lessig, Michael Maier-Gerber, Linus Magnusson, Zied Ben Bouallègue, Ana Prieto Nemesio, Peter D. Dueben, Andrew Brown, Florian Pappenberger, and Florence Rabier. Aifs – ecmwf’s data-driven forecasting system, 2024
2024
-
[15]
Yinglong-weather: Ai-based limited area models for forecasting of non-precipitation surface meteorological vari- ables, 2025
Pengbo Xu, Xiaogu Zheng, Tianyan Gao, Yu Wang, Junping Yin, Juan Zhang, Xuanze Zhang, San Luo, Zhonglei Wang, Zhimin Zhang, Xiaoguang Hu, and Xiaoxu Chen. Yinglong-weather: Ai-based limited area models for forecasting of non-precipitation surface meteorological vari- ables, 2025
2025
-
[16]
Diffusion-lam: Proba- bilistic limited area weather forecasting with diffusion, 2025
Erik Larsson, Joel Oskarsson, Tomas Landelius, and Fredrik Lindsten. Diffusion-lam: Proba- bilistic limited area weather forecasting with diffusion, 2025
2025
-
[17]
Building machine learning limited area models: Kilometer-scale weather forecasting in realistic settings, 2025
Simon Adamov, Joel Oskarsson, Leif Denby, Tomas Landelius, Kasper Hintz, Simon Chris- tiansen, Irene Schicker, Carlos Osuna, Fredrik Lindsten, Oliver Fuhrer, and Sebastian Schemm. Building machine learning limited area models: Kilometer-scale weather forecasting in realistic settings, 2025
2025
-
[18]
Efficient localized adaptation of neural weather forecasting: A case study in the MENA region, 2024
Muhammad Akhtar Munir, Fahad Shahbaz Khan, and Salman Khan. Efficient localized adaptation of neural weather forecasting: A case study in the MENA region, 2024
2024
-
[19]
R. J. H. Dunn, K. M. Willett, P. W. Thorne, E. V . Woolley, I. Durre, A. Dai, D. E. Parker, and R. S. V ose. Hadisd: a quality-controlled global synoptic report database for selected variables at long-term stations from 1973–2011.Climate of the Past, 8(5):1649–1679, October 2012. Data are available without charge under the UK Met Office Non-Commercial Gov...
1973
-
[20]
Dowell, Curtis R
David C. Dowell, Curtis R. Alexander, Eric P. James, Stephen S. Weygandt, Stanley G. Benjamin, Geoffrey S. Manikin, Benjamin T. Blake, John M. Brown, Joseph B. Olson, Ming Hu, Tatiana G. Smirnova, Terra Ladwig, Jaymes S. Kenyon, Ravan Ahmadov, David D. Turner, Jeffrey D. Duda, and Trevor I. Alcott. The high-resolution rapid refresh (hrrr): An hourly updat...
2022
-
[21]
Challenges and design choices for global weather and climate models based on machine learning.Geoscientific Model Development, 11(10):3999–4009, 2018
Peter D Dueben and Peter Bauer. Challenges and design choices for global weather and climate models based on machine learning.Geoscientific Model Development, 11(10):3999–4009, 2018
2018
-
[22]
Can machines learn to predict weather? using deep learning to predict gridded 500-hpa geopotential height from historical weather data
Jonathan A Weyn, Dale R Durran, and Rich Caruana. Can machines learn to predict weather? using deep learning to predict gridded 500-hpa geopotential height from historical weather data. Journal of Advances in Modeling Earth Systems, 11(8):2680–2693, 2019
2019
-
[23]
Improving data-driven global weather prediction using deep convolutional neural networks on a cubed sphere.Journal of Advances in Modeling Earth Systems, 12(9):e2020MS002109, 2020
Jonathan A Weyn, Dale R Durran, and Rich Caruana. Improving data-driven global weather prediction using deep convolutional neural networks on a cubed sphere.Journal of Advances in Modeling Earth Systems, 12(9):e2020MS002109, 2020
2020
-
[24]
The graph neural network model.IEEE Transactions on Neural Networks, 2008
Francesco Scarselli, Marco Gori, Ah Chung Tsoi, Mauro Hagenbuchner, and Gabriele Monfar- dini. The graph neural network model.IEEE Transactions on Neural Networks, 2008
2008
-
[25]
Attention is all you need
Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. InAdvances in Neural Informa- tion Processing Systems, 2017
2017
-
[26]
Kaifeng Bi, Lingxi Xie, Hengheng Zhang, Xin Chen, Xiaotao Gu, and Qi Tian. Pangu-weather: A 3d high-resolution model for fast and accurate global weather forecast.arXiv preprint arXiv:2211.02556, 2022. 11
Pith/arXiv arXiv 2022
-
[27]
Climax: A foundation model for weather and climate.arXiv preprint arXiv:2301.10343, 2023
Tung Nguyen, Johannes Brandstetter, Ashish Kapoor, Jayesh K Gupta, and Aditya Grover. Climax: A foundation model for weather and climate.arXiv preprint arXiv:2301.10343, 2023
Pith/arXiv arXiv 2023
-
[28]
Graph-based neural weather prediction for limited area modeling, 2023
Joel Oskarsson, Tomas Landelius, and Fredrik Lindsten. Graph-based neural weather prediction for limited area modeling, 2023
2023
-
[29]
Regional data-driven weather modeling with a global stretched-grid, 2024
Thomas Nils Nipen, Håvard Homleid Haugen, Magnus Sikora Ingstad, Even Marius Nordhagen, Aram Farhad Shafiq Salihi, Paulina Tedesco, Ivar Ambjørn Seierstad, Jørn Kristiansen, Simon Lang, Mihai Alexe, Jesper Dramsch, Baudouin Raoult, Gert Mertes, and Matthew Chantry. Regional data-driven weather modeling with a global stretched-grid, 2024
2024
-
[30]
Overview and meteorological validation of the wind integration national dataset toolkit
Caroline Draxl, Brian Hodge, James McCaa, and Andrew Clifton. Overview and meteorological validation of the wind integration national dataset toolkit. https://www.nrel.gov/grid/wind- toolkit, 2015. Released under Creative Commons Attribution 3.0 United States (CC BY 3.0 US)
2015
-
[31]
Service change notice 22- 68: Correction of the specification of elevation of the great lakes in the high resolution rapid re- fresh (hrrr)
National Weather Service and NCEP Environmental Modeling Center. Service change notice 22- 68: Correction of the specification of elevation of the great lakes in the high resolution rapid re- fresh (hrrr). https://www.weather.gov/media/notification/pdf2/scn22-68_hrrr_ great_lakes.pdf, 2022. Issued July 19, 2022
2022
-
[32]
Era5 hourly data on single levels from 1940 to present
Copernicus Climate Data Store et al. Era5 hourly data on single levels from 1940 to present. Copernicus Climate Change Service (C3S), 2023
1940
-
[33]
The era5 global reanalysis.Quarterly journal of the royal meteorological society, 146(730):1999–2049, 2020
Hans Hersbach, Bill Bell, Paul Berrisford, Shoji Hirahara, András Horányi, Joaquín Muñoz- Sabater, Julien Nicolas, Carole Peubey, Raluca Radu, Dinand Schepers, et al. The era5 global reanalysis.Quarterly journal of the royal meteorological society, 146(730):1999–2049, 2020
1999
-
[34]
Accelerating the super-resolution convolu- tional neural network
Chao Dong, Chen Change Loy, and Xiaoou Tang. Accelerating the super-resolution convolu- tional neural network. InEuropean conference on computer vision, pages 391–407. Springer, 2016
2016
-
[35]
Photo-realistic single image super-resolution using a generative adversarial network
Christian Ledig, Lucas Theis, Ferenc Huszár, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, et al. Photo-realistic single image super-resolution using a generative adversarial network. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4681–4690, 2017
2017
-
[36]
Fast-srgan
Hasnain Raz. Fast-srgan. https://github.com/HasnainRaz/Fast-SRGAN, 2019. Ac- cessed: 19 January 2026
2019
-
[37]
Kilometer-scale convection allowing model emulation using generative diffusion modeling, 2024
Jaideep Pathak, Yair Cohen, Piyush Garg, Peter Harrington, Noah Brenowitz, Dale Durran, Morteza Mardani, Arash Vahdat, Shaoming Xu, Karthik Kashinath, and Michael Pritchard. Kilometer-scale convection allowing model emulation using generative diffusion modeling, 2024
2024
-
[38]
First lessons learnt from metop
Schlüssel, Peter. First lessons learnt from metop. https://www.ecmwf.int/sites/ default/files/elibrary/2007/15627-first-lessons-learnt-metop.pdf , Septem- ber 2007. Presentation slides, 3–7 September 2007; Accessed: 17 January 2026
2007
-
[39]
Era5: Data documentation – observations
ECMWF. Era5: Data documentation – observations. https:// confluence.ecmwf.int/display/CKB/ERA5%3A+data+documentation#ERA5: datadocumentation-Observations, 2025. Accessed: 17 January 2026
2025
-
[40]
Esa hands over control of the metop-b weather satel- lite to eumetsat
European Space Agency (ESA). Esa hands over control of the metop-b weather satel- lite to eumetsat. https://www.esa.int/Applications/Observing_the_Earth/ Meteorological_missions/MetOp/ESA_hands_over_control_of_the_Metop-B_ weather_satellite_to_EUMETSAT, September 2012. Published: 21 September 2012; Accessed: 17 January 2026
2012
-
[41]
Danra: The kilometer-scale danish regional atmospheric reanalysis, 2025
Xiaohua Yang, Carlos Peralta, Bjarne Amstrup, Kasper Stener Hintz, Søren Borg Thorsen, Leif Denby, Simon Kamuk Christiansen, Hauke Schulz, Sebastian Pelt, and Mathias Schreiner. Danra: The kilometer-scale danish regional atmospheric reanalysis, 2025. Dataset distributed under CC BY 4.0 license. 12 A Regridding - detailed description Regridding must be phy...
2025
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.