Abstract
Most operational climate services providers base their seasonal predictions on initialised general circulation models (GCMs) or empirical statistical techniques. GCMs are widely used but require substantial computational resources, limiting their capacity. In contrast, statistical methods often lack robustness due to the short historical records available. Recent works propose machine learning methods trained on climate model output, leveraging larger sample sizes. Yet, many of these studies focus on prediction tasks that may be restricted in spatial or temporal extent, thereby creating a gap with existing operational predictions. Others fail to disentangle the sources of skill in the context of climate change, where strong trends provide spurious estimates. This study combines variational inference with transformers to predict global and regional seasonal anomalies of temperature and rainfall. The model is trained on output from CMIP6 and tested using ERA5 reanalysis data. Temperature predictions demonstrate skill beyond the climatology and climate-change trend and even outperform the numerical state-of-the-art system SEAS5 in some ocean and land areas. Precipitation forecasts show more limited skill, with fewer regions outperforming climatology and fewer surpassing SEAS5. Furthermore, the consistency found in both teleconnections and skill spatial patterns against SEAS5 suggests that both systems build on similar sources of predictability.
Similar content being viewed by others
Data availability
All the data used are publicly available or restricted to the signed-up users. SEAS5 and ERA5 data were downloaded from the official website of Copernicus Climate Data (CDS) at https://cds.climate.copernicus.eu/. CMIP6 datasets were downloaded from the Earth System Grid Federation (ESGF). CMIP6 and ERA5 datasets were pre-processed using ESMValTool (https://esmvaltool.org).
Code availability
The code used for data processing, model training, inference, and evaluation is available at https://doi.org/10.5281/zenodo.18172866.
References
Lorenz, E. N. Deterministic Nonperiodic Flow. https://journals.ametsoc.org/view/journals/atsc/20/2/1520-0469_1963_020_0130_dnf_2_0_co_2.xml (1963).
Palmer, T. N. & Anderson, D. L. T. The prospects for seasonal forecasting—a review paper. Q. J. R. Meteorol. Soc. 120, 755–793 (1994).
Challinor, A. J., Slingo, J. M., Wheeler, T. R. & Doblas-Reyes, F. J. Probabilistic simulations of crop yield over western India using the DEMETER seasonal hindcast ensembles. Tellus Dyn. Meteorol. Oceanogr. 57, 498–512 (2005).
Pérez-Zanón, N. et al. Lessons learned from the co-development of operational climate forecast services for vineyards management. Clim. Serv. 36, 100513 (2024).
García-Morales, M. B. & Dubus, L. Forecasting precipitation for hydroelectric power management: how to exploit GCM’s seasonal ensemble forecasts. Int. J. Climatol. 27, 1691–1705 (2007).
Lledó, L., Cionni, I., Torralba, V., Bretonnière, P.-A. & Samsó, M. Seasonal prediction of Euro-Atlantic teleconnections from multiple systems. Environ. Res. Lett. 15, 074009 (2020).
Ramon, J., Lledó, L., Bretonnière, P.-A., Samsó, M. & Doblas-Reyes, F. J. A perfect prognosis downscaling methodology for seasonal prediction of local-scale wind speeds. Environ. Res. Lett. https://doi.org/10.1088/1748-9326/abe491 (2021).
Thomson, M. C. et al. Malaria early warnings based on seasonal climate forecasts from multi-model ensembles. Nature 439, 576–579 (2006).
Saha, S. et al. The NCEP climate forecast system version 2. J. Clim. 27, 2185–2208 (2014).
Johnson, S. J. et al. SEAS5: The new ECMWF seasonal forecast system. Geosci. Model Dev. 12, 1087–1117 (2019).
Doblas-Reyes, F. J., García-Serrano, J., Lienert, F., Biescas, A. P. & Rodrigues, L. R. L. Seasonal climate predictability and forecasting: Status and prospects. Wiley Interdiscip. Rev. Clim. Change 4, 245–268 (2013).
Meehl, G. A. et al. Initialized Earth System prediction from subseasonal to decadal timescales. Nat. Rev. Earth Environ. 2, 340–357 (2021).
Scaife, A. A. et al. Does increased atmospheric resolution improve seasonal climate predictions? Atmos. Sci. Lett. 20, e922 (2019).
Materia, S. et al. Impact of atmosphere and land surface initial conditions on seasonal forecasts of global surface temperature. J. Clim. https://doi.org/10.1175/JCLI-D-14-00163.1 (2014).
Ardilouze, C. et al. Multi-model assessment of the impact of soil moisture initialization on mid-latitude summer predictability. Clim. Dyn. 49, 3959–3974 (2017).
Weisheimer, A. & Palmer, T. N. On the reliability of seasonal climate forecasts. J. R. Soc. Interface 11, 20131162 (2014).
Eden, J. M., van Oldenborgh, G. J., Hawkins, E. & Suckling, E. B. A global empirical system for probabilistic seasonal climate prediction. Geosci. Model Dev. 8, 3947–3973 (2015).
Hao, Z., Singh, V. P. & Xia, Y. Seasonal drought prediction: advances, challenges, and future prospects. Rev. Geophys. 56, 108–141 (2018).
Andersson, T. R. et al. Seasonal Arctic sea ice forecasting with probabilistic deep learning. Nat. Commun. 12, 5124 (2021).
Gibson, P. B. et al. Training machine learning models on climate model output yields skillful interpretable seasonal precipitation forecasts. Commun. Earth Environ. 2, 159 (2021).
Materia, S. et al. Artificial intelligence for climate prediction of extremes: state of the art, challenges, and future perspectives. WIREs Clim. Change https://doi.org/10.1002/wcc.914 (2024).
Bi, K. et al. Accurate medium-range global weather forecasting with 3D neural networks. Nature 619, 533–538 (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).
Lam, R. et al. Learning skillful medium-range global weather forecasting. Science 382, 1416–1421 (2023).
Price, I. et al. Probabilistic weather forecasting with machine learning. Nature 637, 84–90 (2025).
Kochkov, D. et al. Neural general circulation models for weather and climate. Nature 632, 1060–1066 (2024).
Ham, Y.-G., Kim, J.-H. & Luo, J.-J. Deep learning for multi-year ENSO forecasts. Nature 573, 568–572 (2019).
Felsche, E. & Ludwig, R. Applying machine learning for drought prediction in a perfect model framework using data from a large ensemble of climate simulations. Nat. Hazards Earth Syst. Sci. 21, 3679–3691 (2021).
Ding, H., Newman, M., Alexander, M. A. & Wittenberg, A. T. Skillful climate forecasts of the tropical Indo-Pacific Ocean using model-analogs. J. Clim. https://doi.org/10.1175/JCLI-D-17-0661.1 (2018).
Mahmood, R. et al. Constraining low-frequency variability in climate projections to predict climate on decadal to multi-decadal timescales—a poor man’s initialized prediction system. Earth Syst. Dyn. 13, 1437–1450 (2022).
Donat, M. G., Mahmood, R., Cos, P., Ortega, P. & Doblas-Reyes, F. Improving the forecast quality of near-term climate projections by constraining internal variability based on decadal predictions and observations. Environ. Res. Clim. 3, 035013 (2024).
Cos, P., Marcos-Matamoros, R., Donat, M., Mahmood, R. & Doblas-Reyes, F. J. Near-term Mediterranean summer temperature climate projections: a comparison of constraining methods. J. Clim. 37, 4367–4388 (2024).
Rader, J. K. & Barnes, E. A. Optimizing seasonal-to-decadal analog forecasts with a learned spatially-weighted mask. Geophys. Res. Lett. 50, e2023GL104983 (2023).
Belkin, M., Hsu, D., Ma, S. & Mandal, S. Reconciling modern machine learning practice and the bias-variance trade-off. Proc. Natl. Acad. Sci. 116, 15849–15854 (2019).
Curth, A. Classical statistical (in-sample) intuitions don’t generalize well: a note on bias-variance tradeoffs, overfitting and moving from fixed to random designs. Preprint at http://arxiv.org/abs/2409.18842 (2024).
Ling, F. et al. Multi-task machine learning improves multi-seasonal prediction of the Indian Ocean Dipole. Nat. Commun. 13, 7681 (2022).
Beobide-Arsuaga, G., Düsterhus, A., Müller, W. A., Barnes, E. A. & Baehr, J. Spring regional sea surface temperatures as a precursor of European summer heatwaves. Geophys. Res. Lett. 50, e2022GL100727 (2023).
Pan, B. et al. Improving seasonal forecast using probabilistic deep learning. J. Adv. Model. Earth Syst. 14, e2021MS002766 (2022).
Patterson, M., Weisheimer, A., Befort, D. J. & O’Reilly, C. H. The strong role of external forcing in seasonal forecasts of European summer temperature. Environ. Res. Lett. 17, 104033 (2022).
Tippett, M. K. & Becker, E. J. Trends, skill, and sources of skill in initialized climate forecasts of global mean temperature. Geophys. Res. Lett. 51, e2024GL110703 (2024).
Goddard, L. et al. A verification framework for interannual-to-decadal predictions experiments. Clim. Dyn. 40, 245–272 (2013).
Risbey, J. S. et al. Common issues in verification of climate forecasts and projections. Climate 10, 83 (2022).
Kingma, D. P. & Welling, M. Auto-encoding variational Bayes. Proc. 2nd International Conference on Learning Representations (ICLR) (2014).
Kingma, D. P. & Welling, M. An introduction to variational autoencoders. https://doi.org/10.48550/ARXIV.1906.02691 (2019).
Dosovitskiy, A. et al. An image is worth 16x16 words: Transformers for image recognition at scale. Proc. 9th International Conference on Learning Representations (ICLR) (2021).
LeCun, Y., Bengio, Y. & Hinton, G. Deep learning. Nature 521, 436–444 (2015).
Eyring, V. et al. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev. 9, 1937–1958 (2016).
Hersbach, H. et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 146, 1999–2049 (2020).
Kim, H.-M., Webster, P. J. & Curry, J. A. Seasonal prediction skill of ECMWF System 4 and NCEP CFSv2 retrospective forecast for the Northern Hemisphere Winter. Clim. Dyn. 39, 2957–2973 (2012).
Kumar, A., Chen, M. & Wang, W. Understanding prediction skill of seasonal mean precipitation over the tropics. https://doi.org/10.1175/JCLI-D-12-00731.1 (2013).
Risbey, J. S. et al. Standard assessments of climate forecast skill can be misleading. Nat. Commun. 12, 4346 (2021).
Fisher, R. A. Frequency distribution of the values of the correlation coefficient in samples from an indefinitely large population. Biometrika 10, 507–521 (1915).
Iturbide, M. et al. An update of IPCC climate reference regions for subcontinental analysis of climate model data: definition and aggregated datasets. Earth Syst. Sci. Data 12, 2959–2970 (2020).
Yang, Z. & Villarini, G. Examining the capability of reanalyses in capturing the temporal clustering of heavy precipitation across Europe. Clim. Dyn. 53, 1845–1857 (2019).
Kent, C. et al. Skilful global seasonal predictions from a machine learning weather model trained on reanalysis data. NPJ Clim. Atmos. Sci. 8, 314 (2025).
Watt-Meyer, O. et al. ACE2: accurately learning subseasonal to decadal atmospheric variability and forced responses. NPJ Clim. Atmos. Sci. 8, 205 (2025).
MacLachlan, C. et al. Global Seasonal forecast system version 5 (GloSea5): a high-resolution seasonal forecast system. Q. J. R. Meteorol. Soc. https://doi.org/10.1002/qj.2396 (2015).
Kolloff, C. et al. Minimum-excess-work guidance. Preprint at https://doi.org/10.48550/arXiv.2505.13375 (2025).
Rackow, T. et al. Multi-year simulations at kilometre scale with the Integrated Forecasting System coupled to FESOM2.5 and NEMOv3.4. Geosci. Model Dev. 18, 33–69 (2025).
Song, Y. et al. Score-based generative modeling through stochastic differential equations. Proc. 9th International Conference on Learning Representations (ICLR) (2021).
Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models. Adv. Neural Inf. Process. Syst. 33, 6840–6851 (2020).
Lipman, Y. et al. Flow matching for generative modeling. Proc. 11th International Conference on Learning Representations (ICLR) (2023).
Murphy, K. P. Probabilistic machine learning: advanced topics.
Wilks, D. S. Statistical Methods in the Atmospheric Sciences: An Introduction (Elsevier, 2019).
Vaswani, A. et al. Attention is all you need. Adv. Neural Inf. Process. Syst. 30, 6000–6010 (2017).
Bodnar, C. et al. A foundation model for the Earth system. Nature 641, 1180–1187 (2025).
Nguyen, T. et al. Scaling transformer neural networks for skillful and reliable medium-range weather forecasting. Adv. Neural Inf. Process. Syst. 37, 68740–68771 (2024).
Nguyen, T. et al. ClimaX: A foundation model for weather and climate. Proc. 40th International Conference on Machine Learning (ICML), 26141–26158 (2023).
Righi, M. et al. Earth System Model Evaluation Tool (ESMValTool) v2.0—technical overview. Geosci. Model Dev. 13, 1179–1199 (2020).
Wulff, C. O., Vitart, F. & Domeisen, D. I. V. Influence of trends on subseasonal temperature prediction skill. Q. J. R. Meteorol. Soc. 148, 1280–1299 (2022).
Mahlstein, I., Spirig, C., Liniger, M. A. & Appenzeller, C. Estimating daily climatologies for climate indices derived from climate model data and observations. J. Geophys. Res.: Atmos. https://doi.org/10.1002/2014JD022327 (2015).
Murphy, Alan. Skill scores based on the mean square error and their relationships to the correlation coefficient. Mon. Weather Rev. 116, 2417–2424 (1988).
Zhao, S., Song, J. & Ermon, S. InfoVAE: Balancing learning and inference in variational autoencoders. Proc. 33rd AAAI Conf. Artif. Intell. 33, 5885–5892 (2019).
Acknowledgements
This work was supported by the AI4Drought (ESA AI4SCIENCE; contract number 4000137110/22/I-EF) and CERISE (European Union; grant agreement No. 101082139) projects. A.D. holds a fellowship within the “Generación D” initiative, Red.es, Ministerio para la Transformación Digital y de la Función Pública, for talent attraction (C005/24-ED CV1), funded by the European Union NextGenerationEU funds, through PRTR. M.G.D. and S.M. are grateful for support from the Horizon Europe project EXPECT (Grant 101137656). The authors thank Pierre-Antoine Bretonnier and Margarida Samsó for their assistance in downloading and formatting part of the data.
Author information
Authors and Affiliations
Contributions
Ll.P., A.D., and A.P. conceived the research idea. Ll.P. and A.P. implemented the deep learning code. Ll.P. implemented the validation pipeline. D.C. contributed to the development of the code. Ll.P. drafted the manuscript with input from all co-authors. All authors discussed the results and revised the manuscript. M.D., S.M., A.M., J.P., L.R., and A.S. supervised the project.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
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
Palma, L., Peraza, A., Civantos-Prieto, D. et al. Data-driven seasonal climate predictions via variational inference and transformers. npj Clim Atmos Sci (2026). https://doi.org/10.1038/s41612-026-01320-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41612-026-01320-z
