Abstract
Decadal climate prediction offers critical scientific guidance for policymakers and supports socioeconomic sustainability. Nevertheless, state-of-the-art dynamical models remain limited in predicting the intricate multidecadal variability in Eurasian surface air temperature (SAT) at mid-to-high latitudes several years in advance. Here, we present a hybrid deep learning model–Gated Recurrent Unit augmented with batch normalization and attention mechanism (GRUBA) –that significantly enhances decadal prediction skill through advanced postprocessing of multi-model ensemble outputs. By incorporating K-means temporal clustering, GRUBA sequentially refines each SAT cluster according to its distinct decadal variability. During the test period of 2004–2021, the averaged anomaly correlation coefficient skill improves from –0.23 to 0.83, and the mean square skill score from –0.37 to 0.68 compared to the unweighted multi-model ensemble mean. This substantial improvement arises in part from the attention mechanism, which effectively reduces model spread. Furthermore, SHAP (Shapley Additive exPlanations) analysis reveals that GRUBA assigns higher weights to better-performing ensemble members.
Similar content being viewed by others
Data availability
The Surface elevation data from the NGDC under the NOAA is available from https://www.ngdc.noaa.gov/mgg/global/relief/ETOPO2/ETOPO2v2-2006/ETOPO2v2c. The T2m monthly averaged reanalysis data from ECMWF (ERA5) is available from [https://doi.org/10.24381/cds.f17050d7]. The model data from the Component A hindcast experiment of CMIP6 DCPP (dcppA-hindcast) is available from [https://esgf-node.llnl.gov/ search/cmip6/](https://esgf-node.llnl.gov/%20search/cmip6).
References
Meehl, G. A. et al. Decadal prediction. Bull. Am. Meteorol. Soc. 90, 1467–1486 (2009).
Oldenborgh, G. J. et al. Decadal prediction skill in a multi-model ensemble. Clim. Dyn. 38, 1263–1280 (2012).
Bellucci, A. et al. An assessment of a multi-model ensemble of decadal climate predictions. Clim. Dyn. 44, 2787–2806 (2015).
Boer, G. J. et al. The Decadal Climate Prediction Project (DCPP) contribution to CMIP6. Geosci. Model Dev. 9, 3751–3777 (2016).
Wu, B. & Xin, X. Short commentary on CMIP6 Decadal Climate Prediction Project (DCPP). Clim. Change Res. 15, 476–480 (2019).
Hermanson, L. et al. WMO global annual to decadal climate update: a prediction for 2021–25. Bull. Am. Meteorol. Soc. 103, E1117–E1129 (2022).
Yeager, S. G. Predicting near-term changes in the earth system: a large ensemble of initialized decadal prediction simulations using the community earth system model. Bull. Am. Meteor. Soc. 99, 1867–1886 (2018).
Smith, D. M. et al. Robust skill of decadal climate predictions. npj Clim. Atmos. Sci. 2, 13 (2019).
Guemas, V. et al. The Indian Ocean: the region of highest skill worldwide in decadal climate prediction. J. Clim. 26, 726–739 (2013).
Meehl, G. A., Hu, A. & Teng, H. Initialized decadal prediction for transition to positive phase of the Interdecadal Pacific Oscillation[J]. Nat. Commun. 7, 11718 (2016).
Saurral, R. I. et al. Decadal predictability and prediction skill of sea surface temperatures in the South Pacific region. Clim. Dyn. 4, 3945–3958 (2020).
Borchert, L. F. et al. Improved decadal predictions of North Atlantic subpolar gyre SST in CMIP6. Geophys. Res. Lett. 48, e2020GL091307 (2021).
Han, T. et al. Projected changes in temperature and precipitation extremes over the Silk Road Economic Belt regions by the Coupled Model Intercomparison Project Phase 5 multi-model ensembles. Int. J. Climatol. 38, 4077–4091 (2018).
Zhao, Y. et al. Extreme temperature indices in Eurasia in a CMIP6 multi-model ensemble: Evaluation and projection. Int. J. Climatol. 41, 5368–5385 (2021).
Monerie, P. A. et al. A role of the Atlantic Ocean in predicting summer surface air temperature over North East Asia?. Clim. Dyn. 51, 473–491 (2018).
Sgubin, G. et al. Systematic investigation of skill opportunities in decadal prediction of air temperature over Europe. Clim. Dyn. 57, 3245–3263 (2021).
Hu, S., Zhou, T. & Wu, B. Accelerated warming of high mountain Asia predicted at multiple years ahead. Sci. Bull. 70, 419–428 (2025).
Huang, Y. Y., Huang, N. & Zhao, Q. Decadal prediction skill for Eurasian surface air temperature in CMIP6 models. Atmos. Ocean. Sci. Lett. 17, 100377 (2024).
Cohen & Coauthors, J. Recent Arctic amplification and extreme mid-latitude weather. Nat. Geosci. 7, 627–637 (2014).
Shepherd, T. G. Effects of a warming Arctic. Science 353, 989–990 (2016).
Li, S. L. & Luo, F. F. Lead-lag connection of the Atlantic Multidecadal Oscillation (AMO) with East Asian surface air temperatures in instrumental records. Atmos. Ocean. Sci. Lett. 6, 138–143 (2013).
Jin, C. H., Wang, B., Yang, Y.-M. & Liu, J. “Warm Arctic-cold Siberia” as an internal mode instigated by North Atlantic warming. Geophys. Res. Lett. 47, e2019GL086248 (2020).
Luo, B. H. et al. Decadal variability of winter warm Arctic-cold Eurasia dipole patterns modulated by Pacific Decadal Oscillation and Atlantic Multidecadal Oscillation. Earth’s Future 10, e2021EF002351 (2022).
Dai, A., Fyfe, J. C., Xie, S.-P. & Dai, X. Decadal modulation of global surface temperature by internal climate variability. Nat. Clim. Change 5, 555–559 (2015).
Luo, D. H. et al. Winter Eurasian cooling linked with the Atlantic Multidecadal Oscillation. Environ. Res. Lett. 12, 125002 (2017).
Outten, S., Davy, R. & Esau, I. Eurasian winter cooling: Intercomparison of reanalyses and CMIP5 data sets. Atmos. Ocean. Sci. Lett. 6, 324–331 (2013).
Outten, S. et al. Reconciling conflicting evidence for the cause of the observed early 21st century Eurasian cooling. Weather Clim. Dyn. 4, 95–114 (2023).
Zhang, P. F. et al. A stratospheric pathway linking a colder Siberia to Barents-Kara Sea sea ice loss. Sci. Adv. 4, eaat6025 (2018).
Hu, S. & Zhou, T. J. Skillful prediction of summer rainfall in the Tibetan Plateau on multiyear time scales. Sci. Adv. 7, eabf9395 (2021).
Eade, R. et al. Do seasonal–to–decadal climate predictions underestimate the predictability of the real world?. Geophys. Res. Lett. 41, 5620–5628 (2014).
Smith, D. M. et al. North Atlantic climate far more predictable than models imply. Nature 583, 796–800 (2020).
Boer, G. J., Merryfield, W. J. & Kharin, V. V. Relationships between potential, attainable, and actual skill in a decadal prediction experiment. Clim. Dyn. 52, 4813–4831 (2019).
Sansom, P. G. et al. Bayesian framework for verification and recalibration of ensemble forecasts: How uncertain is NAO predictability?. J. Clim. 29, 995–1012 (2016).
Haupt, S. E. et al. Towards implementing artificial intelligence post-processing in weather and climate: Proposed actions from the Oxford 2019 workshop. Philos. Trans. R. Soc. A. 379, 20200091 (2021).
Cho, K. et al. Learning phrase representations using RNN encoder–decoder for statistical machine translation. In Proc. 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). 1724–1734 (Association for Computational Linguistics, 2014).
Uluocak, I. & Bilgili, M. Daily air temperature forecasting using LSTM–CNN and GRU–CNN models. Acta Geophys. 72, 2107–2126 (2024).
Zhang, Z. et al. Monthly and quarterly sea surface temperature prediction based on gated recurrent unit neural network. J. Mar. Sci. Eng. 8, 249 (2020).
Loffe, S. & Szegedy, C. Batch normalization: accelerating deep network training by reducing internal covariate shift. In Proc. of the 32nd International Conference on International Conference on Machine Learning (ICML’15). 37, 448–456 (JMLR.org, 2015).
Vaswani, A. et al. Attention is all you need. In Proc. of the 31st International Conference on Neural Information Processing Systems (NIPS’17). 30, 5998–6008 (Curran Associates Inc, 2017).
Lundberg, S. M. & Lee, Su-In. A unified approach to interpreting model predictions. In Proc. of the 31st International Conference on Neural Information Processing Systems (NIPS’17). 4768–4777 (Curran Associates Inc, 2017).
Lu, Z. Y. et al. Regional forecast of heavy precipitation and interpretability based on TD–VAE. In 2021 40th Chinese Control Conference (CCC). IEEE. 7260–7265 (2017).
Dutta, D. & Sankar, K. P. Interpretation of black box for short–term predictions of pre–monsoon cumulonimbus cloud events over Kolkata. J. Data Inf. Manag. 4, 167–183 (2022).
Griffin, S. M., Wimmers, A. & Velden, C. S. Predicting rapid intensification in north Atlantic and eastern north pacific tropical cyclones using a convolutional neural network. Weather Forecast. 37, 1333–1355 (2022).
Bai, S., Kolter, J. Z. & Koltun, V. Convolutional sequence modeling revisited. In 6th International Conference on Learning Representations (ICLR) (OpenReview.net, 2018).
Zeng, A. et al. Are transformers effective for time series forecasting? In Proc. AAAI Conference on Artificial Intelligence Vol 37, 11121–11128 (2023).
Santurkar, S. et al. A. How does batch normalization help optimization?. Adv. Neural Inf. Process. Syst. 31, 2488–2498 (2018).
Hersbach, H. et al. ERA5 monthly averaged data on single levels from 1940 to present. In Copernicus Climate Change Service (C3S) Climate Data Store (CDS). https://doi.org/10.24381/cds.f17050d7 (2023)
Goddard, L. et al. A verification framework for interannual–to–decadal predictions experiments. Clim. Dyn. 40, 245–272 (2013).
Lloyd, S. Least squares quantization in PCM. IEEE Trans. Inf. Theory 28, 29–137 (1982).
Dasgupta, S. & Freund, Y. Random projection trees for vector quantization. Inf. Theory IEEE Trans. 55, 3229–3242 (2009).
Acknowledgements
This research is funded by the National Key Research and Development Program of China [grant number: 2023YFF0806500].
Author information
Authors and Affiliations
Contributions
Y.H. designed the study. Y.C. contributed to the analyses and produced the figures. Y.C. and Y.H. wrote the first draft of the manuscript, which was reviewed by D.Q., D.Z, N.H., Z.Y., and H.W.
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
Chen, Y., Huang, Y., Qian, D. et al. Deep learning model for enhancing decadal prediction of Eurasian surface air temperature. npj Clim Atmos Sci (2026). https://doi.org/10.1038/s41612-026-01337-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41612-026-01337-4
