Skip to content
DraconiticResearch
Search Sources API
203,234 studies31 sourcesUpdated 38 min ago
Peer-reviewedCC BYNature

Operational tropical cyclone forecasting with AI

  • Ferran Alet1
  • Tom R. Andersson2
  • Ilan Price3
  • Stratis Markou4
  • Andrew El-Kadi5
  • Dominic A. Masters5
  • Amy Li6
  • Samier Merchant7
  • Natalie Williams7
  • Gregory Thornton5
  • Ken MacKay5
  • Olivia Graham7
20 more authors
  • Akib Uddin7
  • Ben Gaiarin5
  • Devaja Shah5
  • Elinor Kruse5
  • Wallace A. Hogsett8
  • David Zelinsky8
  • John P. Cangialosi8
  • Jonathan Martinez9,8
  • James Franklin9
  • Mark DeMaria9
  • Kate Musgrave9
  • Caroline L. Bain10
  • Helen A. Titley10
  • Jacklynn Stott5
  • Rémi Lam5
  • Aaron Bell7
  • Paul Komarek5
  • Matthew Willson5
  • Álvaro Sánchez‐González5
  • Peter Battaglia5
10 affiliations
  1. 1Google DeepMind, London, UK. [email protected]
  2. 2Google DeepMind, London, UK. [email protected]
  3. 3Google DeepMind, London, UK. [email protected]
  4. 4Google DeepMind, London, UK. [email protected]
  5. 5Google DeepMind, London, UK
  6. 6University of Waterloo, Waterloo, Ontario, Canada
  7. 7Google Research, Mountain View, CA, USA
  8. 8NOAA/NWS/NCEP National Hurricane Center, Miami, FL, USA
  9. 9Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, CO, USA
  10. 10UK Met Office, Exeter, UK
Open at sourceDOICite

Abstract

Tropical cyclones are among the most dangerous and costly weather phenomena, but forecasting them remains challenging. Here we introduce WeatherNext Cyclones (WN-C), an artificial intelligence (AI)-based operational weather model that produces state-of-the-art ensemble forecasts of the track, intensity and size of tropical cyclones worldwide. Trained on a combination of global analysis data 1 and a global database of historical tropical cyclones 2,3 , WN-C generates large ensembles of possible global weather and cyclone scenarios extending 15 days into the future. When evaluated on tropical cyclones from 2023 to 2025, the track, intensity and wind-radius predictions from WN-C offer an average lead-time advantage of 1 day or more over leading operational models—an improvement in accuracy comparable to the progress seen in the last decade of operational development. We achieved these results using inputs that are orders of magnitude coarser than regional models, suggesting that high resolution is not a strict prerequisite for state-of-the-art intensity forecasting and that these coarse atmospheric data contain more intensity signal than has been previously recognized. Including predictions from WN-C in a weighted-average consensus ensemble improves its skill substantially. The scalability of WN-C enables ensembles of up to 1,000 members, which are better at capturing rare events than conventional 50-member ensembles. By providing advanced operational ensemble guidance to human forecasters, this work represents a step change towards more reliable and timely forecasts and warnings that can help to protect lives and to mitigate the devastating effects of tropical cyclones.

Related

  1. Butterfly Effect Confirmed in Global AI Weather Models: Evidence from Tropical Cyclone Forecasting
    Jeremy Cheuk-Hin Leung, Daosheng Xu, Weiye Yu +9
    arXivPreprintCC BY-NC-SA 4.0Full textAtmospheric science & meteorologyStatistical & nonlinear physicsarXiv
  2. FAST-ML: A Hybrid Physics-Machine Learning Framework for Tropical Cyclone Intensity Forecasting
    Shijie Xiao, Jonathan Lin, Thomas Ehrmann +1
    arXivPreprintarXiv licenceFull textAtmospheric science & meteorologyarXiv
  3. Skillful Seasonal Tropical Cyclone Prediction with AI Weather--Climate Models Forced by Dynamically Predicted SSTs
    Hiroyuki Murakami, Baoqiang Xiang
    EarthArXivPreprintPublisher termsFull textAtmospheric science & meteorologyEarthArXiv
  4. Tibetan Plateau uplift increased western North Pacific tropical cyclone frequency
    Tingyu Zhang, Tianjun Zhou, Linqiang He +3
    Nature CommunicationsPeer-reviewedCC BY-NC-ND 4.0Full textClimate scienceCrossref
  5. Probabilistic assessment of tropical cyclone recovery burden across US Atlantic and Gulf Coast counties
    Simona Meiler, Nikola Blagojevic, Meredith Lochhead +2
    EGUspherePreprintCC BY 4.0Full textGeneral earth scienceEGUsphere

Cite

BibTeXRIS
BibTeX
@article{alet2026operational,
  title = {{Operational tropical cyclone forecasting with AI}},
  author = {Ferran Alet and Tom R. Andersson and Ilan Price and Stratis Markou and Andrew El-Kadi and Dominic A. Masters and Amy Li and Samier Merchant and Natalie Williams and Gregory Thornton and Ken MacKay and Olivia Graham and Akib Uddin and Ben Gaiarin and Devaja Shah and Elinor Kruse and Wallace A. Hogsett and David Zelinsky and John P. Cangialosi and Jonathan Martinez and James Franklin and Mark DeMaria and Kate Musgrave and Caroline L. Bain and Helen A. Titley and Jacklynn Stott and Rémi Lam and Aaron Bell and Paul Komarek and Matthew Willson and Álvaro Sánchez‐González and Peter Battaglia},
  journal = {Nature},
  year = {2026},
  month = aug,
  doi = {10.1038/s41586-026-10953-2},
  url = {https://doi.org/10.1038/s41586-026-10953-2},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Operational tropical cyclone forecasting with AI
AU  - Ferran Alet
AU  - Tom R. Andersson
AU  - Ilan Price
AU  - Stratis Markou
AU  - Andrew El-Kadi
AU  - Dominic A. Masters
AU  - Amy Li
AU  - Samier Merchant
AU  - Natalie Williams
AU  - Gregory Thornton
AU  - Ken MacKay
AU  - Olivia Graham
AU  - Akib Uddin
AU  - Ben Gaiarin
AU  - Devaja Shah
AU  - Elinor Kruse
AU  - Wallace A. Hogsett
AU  - David Zelinsky
AU  - John P. Cangialosi
AU  - Jonathan Martinez
AU  - James Franklin
AU  - Mark DeMaria
AU  - Kate Musgrave
AU  - Caroline L. Bain
AU  - Helen A. Titley
AU  - Jacklynn Stott
AU  - Rémi Lam
AU  - Aaron Bell
AU  - Paul Komarek
AU  - Matthew Willson
AU  - Álvaro Sánchez‐González
AU  - Peter Battaglia
PY  - 2026
DA  - 2026/08/06
JO  - Nature
DO  - 10.1038/s41586-026-10953-2
UR  - https://doi.org/10.1038/s41586-026-10953-2
AB  - Tropical cyclones are among the most dangerous and costly weather phenomena, but forecasting them remains challenging. Here we introduce WeatherNext Cyclones (WN-C), an artificial intelligence (AI)-based operational weather model that produces state-of-the-art ensemble forecasts of the track, intensity and size of tropical cyclones worldwide. Trained on a combination of global analysis data 1 and a global database of historical tropical cyclones 2,3 , WN-C generates large ensembles of possible global weather and cyclone scenarios extending 15 days into the future. When evaluated on tropical cyclones from 2023 to 2025, the track, intensity and wind-radius predictions from WN-C offer an average lead-time advantage of 1 day or more over leading operational models—an improvement in accuracy comparable to the progress seen in the last decade of operational development. We achieved these results using inputs that are orders of magnitude coarser than regional models, suggesting that high resolution is not a strict prerequisite for state-of-the-art intensity forecasting and that these coarse atmospheric data contain more intensity signal than has been previously recognized. Including predictions from WN-C in a weighted-average consensus ensemble improves its skill substantially. The scalability of WN-C enables ensembles of up to 1,000 members, which are better at capturing rare events than conventional 50-member ensembles. By providing advanced operational ensemble guidance to human forecasters, this work represents a step change towards more reliable and timely forecasts and warnings that can help to protect lives and to mitigate the devastating effects of tropical cyclones.
SN  - 0028-0836
SN  - 1476-4687
ER  -