Abstract
Fine-scale near-surface wind field prediction is essential for a wide range of applications. However, most operational and AI-based weather models operate at kilometer-scale resolution, where terrain-induced wind features such as slope jets, flow deflection, and recirculation are systematically averaged out. Here we introduce FuXi-CFD, a machine learning-based framework designed to generate detailed three-dimensional (3D) near-surface wind fields at 30-meter horizontal resolution, using only coarse-resolution atmospheric inputs and high-resolution terrain information. The model is trained on a large-scale dataset generated via computational fluid dynamics (CFD), encompassing a wide range of terrain types and inflow conditions. Although relying only on horizontal wind inputs, FuXi-CFD infers the full 3D wind fields—including latent variables such as vertical velocity and turbulence-related features. It achieves CFD-comparable accuracy while reducing inference time from hours to seconds. Notably, the model also generalizes well to real-world conditions, as demonstrated by consistent performance against independent wind-tower observations. This capability enables real-time wind field reconstruction for terrain-sensitive applications such as wind turbine siting, power forecasting, and wildfire spread modeling.
Similar content being viewed by others
Data availability
The FuXi-CFD dataset generated in this study is publicly available at Zenodo (https://doi.org/10.5281/zenodo.18770845). All data necessary to reproduce the results reported in this paper are publicly available.
Code availability
The inference code and pre-trained model weights used in this study are publicly available at Zenodo (https://doi.org/10.5281/zenodo.18770845).
References
Barthelmie, R. J. & Pryor, S. C. Automated wind turbine wake characterization in complex terrain. Atmos. Meas. Tech. 12, 3463–3484 (2019).
Pan, Y. & Archer, C. L. A hybrid wind-farm parametrization for mesoscale and climate models. Bound. Layer. Meteorol. 168, 469–495 (2018).
Floors, R., Vincent, C. L., Gryning, S.-E., Peña, A. & Batchvarova, E. The wind profile in the coastal boundary layer: wind lidar measurements and numerical modelling. Bound. Layer. Meteorol. 147, 469–491 (2013).
Linn, R., Winterkamp, J., Edminster, C., Colman, J. J. & Smith, W. S. Coupled influences of topography and wind on wildland fire behaviour. Int. J. Wildland Fire 16, 183–195 (2007).
Veers, P. et al. Grand challenges in the science of wind energy. Science 366, eaau2027 (2019).
Clifton, A., Barber, S., Stökl, A., Frank, H. & Karlsson, T. Research challenges and needs for the deployment of wind energy in hilly and mountainous regions. Wind Energy Sci. 7, 2231–2254 (2022).
He, Y., Chan, P. & Li, Q. Wind characteristics over different terrains. J. Wind Eng. Ind. Aerodyn. 120, 51–69 (2013).
Haiden, T. et al. Evaluation of ecmwf forecasts, including the 2018 upgrade. Tech. Rep. ECMWF Technical Memorandum 831. (European Centre for Medium-Range Weather Forecasts, Reading, UK, 2018).
Harris, L., Chen, X., Putman, W., Zhou, L. & Chen, J.-H. A scientific description of the gfdl finite-volume cubed-sphere dynamical core. NOAA Tech. Memo. OAR GFDL 1,109 (2021).
Chen, D. et al. New generation of multi-scale NWP system (GRAPES): General scientific design. Chin. Sci. Bull. 53, 3433–3445 (2008).
Lam, R. et al. Learning skillful medium-range global weather forecasting. Science 382, 1416–1421 (2023).
Pathak, J. et al. Fourcastnet: a global data-driven high-resolution weather model using adaptive Fourier neural operators. Preprint at https://doi.org/10.48550/arXiv.2202.11214 (2022).
Bi, K. et al. Accurate medium-range global weather forecasting with 3d neural networks. Nature 619, 533–538 (2023).
Chen, L. et al. Fuxi: a cascade machine learning forecasting system for 15-day global weather forecast. npj Clim. Atmos. Sci. 6, 190 (2023).
Chen, L. et al. A machine learning model that outperforms conventional global subseasonal forecast models. Nat. Commun. 15, 6425 (2024).
Espeholt, L. et al. Deep learning for twelve hour precipitation forecasts. Nat. Commun. 13, 1–10 (2022).
Price, I. et al. Gencast: diffusion-based ensemble forecasting for medium-range weather. Preprint at https://doi.org/10.48550/arXiv.2312.15796 (2023).
Hersbach, H. et al. The era5 global reanalysis. Q. J. R. Meteorol. Soc. 146, 1999–2049 (2020).
Schoof, J. T. Statistical downscaling in climatology. Geogr. Compass 7, 249–265 (2013).
Martinez-GarcĂa, F. P., Contreras-de Villar, A. & Muñoz-Perez, J. J. Review of wind models at a local scale: advantages and disadvantages. J. Mar. Sci. Eng. 9, 318 (2021).
Soares, P. M. et al. WRF high-resolution dynamical downscaling of era-interim for Portugal. Clim. Dyn. 39, 2497–2522 (2012).
Fernández-González, S. et al. Sensitivity analysis of the wrf model: Wind-resource assessment for complex terrain. J. Appl. Meteorol. Climatol. 57, 733–753 (2018).
Goger, B., Rotach, M. W., Gohm, A., Stiperski, I. & Fuhrer, O. Current challenges for numerical weather prediction in complex terrain: Topography representation and parameterizations. In 2016 international conference on high performance computing & simulation (HPCS), 890–894 (IEEE, 2016).
Dupuy, F., Durand, P. & Hedde, T. Downscaling of surface wind forecasts using convolutional neural networks. Nonlinear Process. Geophys. 30, 553–570 (2023).
Kumar, A. et al. Windsr: Improving spatial resolution of satellite wind speed through super-resolution. IEEE Access 11, 69486–69494 (2023).
Lian, J. et al. Terrawind: a deep learning-based near-surface winds downscaling model for complex terrain region. Geophys. Res. Lett. 51, e2024GL112124 (2024).
Buster, G., Benton, B. N., Glaws, A. & King, R. N. High-resolution meteorology with climate change impacts from global climate model data using generative machine learning. Nat. Energy 9, 894–906 (2024).
Jiang, P. et al. Efficient super-resolution of near-surface climate modeling using the Fourier neural operator. J. Adv. Modeling Earth Syst. 15, e2023MS003800 (2023).
Sinha, S., Benton, B. & Emami, P. On the effectiveness of neural operators at zero-shot weather downscaling. Environ. Data Sci. 4, e21 (2025).
Ding, J.-W. & Hsieh, I.-Y. L. Super-resolution wind mapping with deep learning for scalable renewable energy planning. Commun. Earth Environ. 7, 51 (2025).
Mardani, M. et al. Residual corrective diffusion modeling for km-scale atmospheric downscaling. Commun. Earth Environ. 6, 124 (2025).
Tomasi, E., Franch, G. & Cristoforetti, M. Can AI be enabled to perform dynamical downscaling? a latent diffusion model to mimic kilometer-scale cosmo5. 0_clm9 simulations. Geosci. Model Dev. 18, 2051–2078 (2025).
Lopez-Gomez, I. et al. Dynamical-generative downscaling of climate model ensembles. Proc. Natl. Acad. Sci. USA 122, e2420288122 (2025).
Le Toumelin, L. et al. Emulating the adaptation of wind fields to complex terrain with deep learning. Artif. Intell. Earth Syst. 2, e220034 (2023).
Ferziger, J. H. & Perić, M.Computational methods for fluid dynamics. 586 (Springer, 2002).
Karniadakis, G. E., Israeli, M. & Orszag, S. A. High-order splitting methods for the incompressible Navier-Stokes equations. J. Comput. Phys. 97, 414–443 (1991).
Meteodyn. Meteodyn WT - Wind Resource Assessment Software. https://www.meteodyn.com/.
WindSim AS. WindSim - CFD software for wind energy. https://www.windsim.com/.
Envision Energy. Greenwich - Wind Farm Design Platform. https://www.envision-group.com/.
Icos research infrastructure: Atmospheric tower observations. https://data.icos-cp.eu (2025).
U.S. Geological Survey. Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global. USGS Earth ResourcesObservation and Science (EROS) Center https://doi.org/10.5066/F7PR7TFT (2014).
Zanaga, D. et al. ESA WorldCover 10 m 2020 v100. Zenodo https://doi.org/10.5281/zenodo.5571936 (2021).
Hasager, C. B. et al. Effective roughness calculated from satellite-derived land cover maps and hedge-information used in a weather forecasting model. Bound.Layer. Meteorol. 109, 227–254 (2003).
Dosovitskiy, A. et al. An image is worth 16x16 words: transformers for image recognition at scale. Preprint at https://doi.org/10.48550/arXiv.2010.11929 (2020).
Acknowledgements
This work was supported by the AI for Science Program of Shanghai Municipal Commission of Economy and Informatization (2025-GZL-RGZN-BTBX-02017, 2025-GZL-RGZN-BTBX-02031) and the Smart Grid National Science and Technology Major Project (2024ZD0800400). Part of this research was performed using the CFFF platform of Fudan University. The authors thank Prof. Dongxiao Zhao (Shanghai Jiao Tong University) for helpful discussions and insightful suggestions.
Author information
Authors and Affiliations
Contributions
C.L. conceived the study, designed the methodology, performed the analyses and visualization, and wrote the original draft of the manuscript. R.T. contributed to AI model training and writing the original draft of the manuscript. S.Y., D.L., X.Z., and Z.H. contributed to validation and manuscript review and editing. H.L. conceived the study, secured funding, supervised the project, provided project administration, and contributed to manuscript review and editing.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Lin, C., Tie, R., Yi, S. et al. Reconstructing fine-scale 3D wind fields with terrain-informed machine learning. Nat Commun (2026). https://doi.org/10.1038/s41467-026-70562-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-026-70562-5
