Deep learning for global sea surface temperature anomaly forecasting with balanced spatiotemporal features

Yan Li; Weiyi Chen; Jia Shi; Zixuan Lin; Yan Li; Weiyi Chen; Jia Shi; Zixuan Lin

doi:10.3934/math.2026919

AIMS Mathematics

2026, Volume 11, Issue 7: 22770-22806. doi: 10.3934/math.2026919

Previous Article Next Article

Research article

Deep learning for global sea surface temperature anomaly forecasting with balanced spatiotemporal features

1.
School of Mathematics and Statistics, Changchun University, Changchun 130012, China
2.
Anyang Wenfeng District Tax Bureau, State Taxation Administration, Anyang 455000, China

Received: 25 April 2026 Revised: 30 June 2026 Accepted: 06 July 2026 Published: 28 July 2026
MSC : 68T07, 62M10, 86A10

With climate change intensifying, sea surface temperature anomalies (SSTAs) have become a key climate indicator. However, their nonlinear, non‑stationary, and periodic characteristics pose challenges to traditional forecasting methods. This paper proposes a hybrid forecasting framework that integrates three-level temporal decomposition with deep learning, extending single-point prediction to multi-region spatiotemporal forecasting. First, a hybrid SCV module integrating singular spectrum analysis (SSA), complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), and variational mode decomposition (VMD) is established to disentangle the original signal into trend, periodic, and stochastic components, which are subsequently reconstructed via fuzzy entropy.These components are then fed into deep learning models (e.g., bidirectional gated recurrent unit (BiGRU)) for prediction. Second, we combine temporal features with spatial features to propose the SCV-spatiotemporal feature coupling block (STFCB) model. This model uses a feature pyramid network (FPN) and an intrinsic mode function (IMF)-guided convolutional block attention module (IMF-guided CBAM) to dynamically fuse spatial information. The results demonstrate that SCV-STFCB significantly outperforms the temporal-only SCV with feature attention (SCV-FA) egret swarm optimization algorithm (ESOA) model, with BiGRU achieving a mean square errora mean square error of 0.000073 and $ R^2 $ of 0.995903, indicating that the proposed spatiotemporal enhancement method improves the test-period predictive performance.
- sea surface temperature anomaly prediction,
- time series decomposition,
- feature pyramid network,
- convolutional block attention module,
- deep learning
Citation: Yan Li, Weiyi Chen, Jia Shi, Zixuan Lin. Deep learning for global sea surface temperature anomaly forecasting with balanced spatiotemporal features[J]. AIMS Mathematics, 2026, 11(7): 22770-22806. doi: 10.3934/math.2026919

Related Papers:

Abstract

With climate change intensifying, sea surface temperature anomalies (SSTAs) have become a key climate indicator. However, their nonlinear, non‑stationary, and periodic characteristics pose challenges to traditional forecasting methods. This paper proposes a hybrid forecasting framework that integrates three-level temporal decomposition with deep learning, extending single-point prediction to multi-region spatiotemporal forecasting. First, a hybrid SCV module integrating singular spectrum analysis (SSA), complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), and variational mode decomposition (VMD) is established to disentangle the original signal into trend, periodic, and stochastic components, which are subsequently reconstructed via fuzzy entropy.These components are then fed into deep learning models (e.g., bidirectional gated recurrent unit (BiGRU)) for prediction. Second, we combine temporal features with spatial features to propose the SCV-spatiotemporal feature coupling block (STFCB) model. This model uses a feature pyramid network (FPN) and an intrinsic mode function (IMF)-guided convolutional block attention module (IMF-guided CBAM) to dynamically fuse spatial information. The results demonstrate that SCV-STFCB significantly outperforms the temporal-only SCV with feature attention (SCV-FA) egret swarm optimization algorithm (ESOA) model, with BiGRU achieving a mean square errora mean square error of 0.000073 and $ R^2 $ of 0.995903, indicating that the proposed spatiotemporal enhancement method improves the test-period predictive performance.

References

[1]	X. F. Wu, J. P. Xu, A review of distribution characteristics, variation modes, and observation methods of upper ocean heat content, J. Mar. Sci., 28 (2010), 46–54.
[2]	IPCC, AR6 Synthesis Report: Climate Change 2023, Geneva, Switzerland: IPCC, 2023.
[3]	F. Pastor, J. A. Valiente, S. Khodayar, A warming Mediterranean: 38 years of increasing sea surface temperature, Remote Sens., 12 (2020), 2687. https://doi.org/10.3390/rs12172687 doi: 10.3390/rs12172687
[4]	Q. X. Lai, W. F. Zhou, X. S. Cui, Y. C. Shi, Interannual variability analysis of sea surface temperature anomalies in major pelagic fishing grounds over the past 40 years, in Chinese, Marine Science, 48 (2024), 33–47.
[5]	C. Frankignoul, Sea surface temperature anomalies, planetary waves, and air-sea feedback in the middle latitudes, Rev. Geophys., 23 (1985), 357–390. https://doi.org/10.1029/RG023i004p00357 doi: 10.1029/RG023i004p00357
[6]	N. A. Rayner, P. Brohan, D. E. Parker, C. K. Folland, J. J. Kennedy, M. Vanicek, et al., Improved analysis of changes and uncertainties in sea surface temperature measured in situ since the mid-nineteenth century: The HadSST2 dataset, J. Climate, 19 (2006), 446–469. https://doi.org/10.1175/JCLI3882.1 doi: 10.1175/JCLI3882.1
[7]	P. J. Minnett, A. Alvera-Azcárate, T. M. Chin, G. K. Corlett, C. L. Gentemann, I. Karagali, et al., Half a century of satellite remote sensing of sea-surface temperature, Remote Sens. Environ., 233 (2019), 111366. https://doi.org/10.1016/j.rse.2019.111366 doi: 10.1016/j.rse.2019.111366
[8]	Y. Zhao, D. Yang, Z. He, C. Liu, R. Hao, J. He, Statistical methods in ocean prediction, In: Global Oceans 2020: Singapore–US Gulf Coast, IEEE, 2020. https://doi.org/10.1109/IEEECONF38699.2020.9389485
[9]	T. Xu, Z. Zhou, Y. Li, C. Wang, Y. Liu, T. Rong, Short-term prediction of global sea surface temperature using deep learning networks, J. Mar. Sci. Eng., 11 (2023), 1352. https://doi.org/10.3390/jmse11071352 doi: 10.3390/jmse11071352
[10]	T. M. Smith, R. W. Reynolds, R. E. Livezey, D. C. Stokes, Reconstruction of historical sea surface temperatures using empirical orthogonal functions, J. Climate, 9 (1996), 1403–1420. https://doi.org/10.1175/1520-0442(1996)009<3C1403:ROHSST>3E2.0.CO;2 doi: 10.1175/1520-0442(1996)009<3C1403:ROHSST>3E2.0.CO;2
[11]	T. M. Smith, R. W. Reynolds, Extended reconstruction of global sea surface temperatures based on COADS data (1854–1997), J. Climate, 16 (2003), 1495–1510. https://doi.org/10.1175/1520-0442(2003)016<3C1495:EROGSS>3E2.0.CO;2 doi: 10.1175/1520-0442(2003)016<3C1495:EROGSS>3E2.0.CO;2
[12]	T. M. Smith, R. W. Reynolds, Improved extended reconstruction of SST (1854–1997), J. Climate, 17 (2004), 2466–2477. https://doi.org/10.1175/1520-0442(2004)017<3C2466:IEROS>3E2.0.CO;2 doi: 10.1175/1520-0442(2004)017<3C2466:IEROS>3E2.0.CO;2
[13]	N. A. Rayner, D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, et al., Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century, J. Geophys. Res., 108 (2003), 4407. https://doi.org/10.1029/2003JD002437 doi: 10.1029/2003JD002437
[14]	I. S. Kang, J. S. Kug, An El NINO prediction system using an intermediate ocean and a statistical atmosphere, Geophys. Res. Lett., 27 (2000), 1167–1170. https://doi.org/10.1029/1999GL011023 doi: 10.1029/1999GL011023
[15]	Y. Kushnir, W. A. Robinson, P. Chang, A. W. Robertson, The physical basis for predicting Atlantic sector seasonal-to-interannual climate variability, J. Climate, 19 (2006), 5949–5970. https://doi.org/10.1175/JCLI3943.1 doi: 10.1175/JCLI3943.1
[16]	T. N. Stockdale, M. A. Balmaseda, A. Vidard, Tropical Atlantic SST prediction with coupled ocean–atmosphere GCMs, J. Climate, 19 (2006), 6047–6061. https://doi.org/10.1175/JCLI3947.1 doi: 10.1175/JCLI3947.1
[17]	W. A. Landman, S. J. Mason, Forecasts of near-global sea surface temperatures using canonical correlation analysis, J. Climate, 14 (2001), 3819–3833. https://doi.org/10.1175/1520-0442(2001)014<3C3819:FONGSS>3E2.0.CO;2 doi: 10.1175/1520-0442(2001)014<3C3819:FONGSS>3E2.0.CO;2
[18]	J. S. Kug, I. S. Kang, J. Y. Lee, J. G. Jhun, A statistical approach to Indian Ocean sea surface temperature prediction using a dynamical ENSO prediction, Geophys. Res. Lett., 31 (2004), L09212. https://doi.org/10.1029/2003GL019209 doi: 10.1029/2003GL019209
[19]	W. Nawi, M. S. Lola, R. Zakariya, N. H. Zainuddinc, A. Aziz K. Abd Hamida, E. Aruchunan, et al., Improved of forecasting sea surface temperature based on hybrid ARIMA and support vector machines models, Mal. J. Fund. Appl. Sci.‌, 17 (2021), 609–620.
[20]	Y. Zhang, Y. C. Tan, F. D. Peng, X. J. Liao, Y. X. Yu, Study on time series prediction model of sea surface temperature based on Ensemble Empirical Mode Decomposition and Autoregressive Integrated Moving Average, J. Mar. Sci., 37 (2019), 9–14. https://doi.org/10.3969/j.issn.1001-909X.2019.01.002 doi: 10.3969/j.issn.1001-909X.2019.01.002
[21]	H. Chen, Research on intelligent prediction method of South China Sea SST based on improved EMD algorithm, Master's Thesis, in Chinese, Harbin Engineering University, 2021. https://doi.org/10.27060/d.cnki.ghbcu.2021.000139
[22]	Y. Han, Y. Z. Cao, L. J. Zhang, R. H. Zhao, C. M. Dong, Spatiotemporal intelligent prediction method for sea surface temperature incorporating IVMD, in Chinese, Hydrographic Surveying and Charting, 44 (2024), 6. https://doi.org/10.3969/j.issn.1671-3044.2024.03.011 doi: 10.3969/j.issn.1671-3044.2024.03.011
[23]	L. W. Li, Spatiotemporal information mining and prediction of the impact of sea surface temperature on precipitation in China based on machine learning, in Chinese, National Marine Environmental Forecasting Center, 2021. https://doi.org/10.27810/d.cnki.ghyhz.2021.000005
[24]	A. Wu, W. W. Hsieh, B. Tang, Neural network forecasts of the tropical Pacific sea surface temperatures, Neur. Networks, 19 (2006), 145–154. https://doi.org/10.1016/j.neunet.2006.01.004 doi: 10.1016/j.neunet.2006.01.004
[25]	F. T. Tangang, W. W. Hsieh, B. Tang, Forecasting the equatorial Pacific sea surface temperatures by neural network models, Climate Dyn., 13 (1997), 135–147. https://doi.org/10.1007/s003820050156 doi: 10.1007/s003820050156
[26]	S. Hochreiter, J. Schmidhuber, Long short-term memory, Neural Comput., 9 (1997), 1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735 doi: 10.1162/neco.1997.9.8.1735
[27]	J. Chung, C. Gulcehre, K. H. Cho, Y. Bengio, Empirical evaluation of gated recurrent neural networks on sequence modeling, arXiv: 1412.3555, 2014. https://doi.org/10.48550/arXiv.1412.3555
[28]	S. Siami-Namini, N. Tavakoli, A. S. Namin, The performance of LSTM and BiLSTM in forecasting time series, In: 2019 IEEE International Conference on Big Data (Big Data). IEEE, 2019, 3285–3292. https://doi.org/10.1109/BigData47090.2019.9005997
[29]	A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, et al., Attention is all you need, In: Advances in Neural Information Processing Systems, 30 (2017).
[30]	D. Tran, L. Bourdev, R. Fergus, L. Torresani, M. Paluri, Learning spatiotemporal features with 3D convolutional networks, In: Proceedings of the IEEE International Conference on Computer Vision, 2015, 4489–4497.
[31]	J. S. Li, X. D. Kuang, Q. Li, Sea surface temperature forecast model based on principal component analysis and LSTM neural network, in Chinese, Marine Forecasts, 40 (2023), 1–10. https://doi.org/10.11737/j.issn.1003-0239.2023.02.001 doi: 10.11737/j.issn.1003-0239.2023.02.001
[32]	Y. Yang, J. Dong, X. Sun, E. Lima, Q. Mu, X. Wang, A CFCC-LSTM model for sea surface temperature prediction, IEEE Geosci. Remote Sens. Lett., 15 (2017), 207–211. https://doi.org/10.1109/LGRS.2017.2777690 doi: 10.1109/LGRS.2017.2777690
[33]	Q. Zhang, H. Wang, J. Dong, G. Zhong, X. Sun, Prediction of sea surface temperature using long short-term memory, IEEE Geosci. Remote Sens. Lett., 14 (2017), 1745–1749. https://doi.org/10.1109/LGRS.2017.2733548 doi: 10.1109/LGRS.2017.2733548
[34]	C. Xiao, N. Chen, C. Hu, K. Wang, J. Gong, Z. Chen, Short and mid-term sea surface temperature prediction using time-series satellite data and LSTM-AdaBoost combination approach, Remote Sens. Environm., 233 (2019), 111358. https://doi.org/10.1016/j.rse.2019.111358 doi: 10.1016/j.rse.2019.111358
[35]	V. Lakshmana Gomathi Nayagam, D. Ponnialagan, S. Jeevaraj, Similarity measure on incomplete imprecise interval information and its applications, Neural Comput. Applic., 32 (2020), 3749–3761. https://doi.org/10.1007/s00521-019-04277-8 doi: 10.1007/s00521-019-04277-8
[36]	B. Qiao, Z. Wu, Z. Tang, G. Wu, Sea surface temperature prediction approach based on 3D CNN and LSTM with attention mechanism, In: Proceedings of the 2022 24th International Conference on Advanced Communication Technology (ICACT), IEEE, 2022,342–347. https://doi.org/10.23919/ICACT53585.2022.9728889
[37]	P. K. Jonnakuti, U. Bhaskar Tata Venkata Sai, A hybrid CNN-LSTM based model for the prediction of sea surface temperature using time-series satellite data, In: EGU General Assembly Conference Abstracts, 2020,817. https://doi.org/10.5194/egusphere-egu2020-817
[38]	Y. Miao, C. Zhang, X. Zhang, L. Zhang, A multivariable convolutional neural network for forecasting synoptic-scale sea surface temperature anomalies in the South China Sea, Weather Forec., 38 (2023), 849–863. https://doi.org/10.1175/WAF-D-22-0094.1 doi: 10.1175/WAF-D-22-0094.1
[39]	J. Liu, L. Wang, F. Hu, P. Xu, D. Zhang, Spatiotemporal fusion prediction of sea surface temperatures based on the graph convolutional neural and long short-term memory networks, Water, 16 (2024), 1725. https://doi.org/10.3390/w16121725 doi: 10.3390/w16121725
[40]	H. Yang, W. Li, S. Hou, J. Guan, S. Zhou, HiGRN: A hierarchical graph recurrent network for global sea surface temperature prediction, ACM Trans. Intell. Syst. Technol., 14 (2023), 73. https://doi.org/10.1145/3597937 doi: 10.1145/3597937
[41]	Y. Z. Cao, Research on spatiotemporal prediction method of sea surface temperature based on graph convolutional neural network, Master's Thesis, in Chinese, Nanjing University of Information Science and Technology, 2024.
[42]	L. N. Wang, P. Ge, H. T. Liu, Y. Song, C. M. Dong HDC-BiGRU sea surface temperature prediction model based on attention mechanism, Mar. Environ. Sci., 41 (2022), 947–956.
[43]	B. Feng, D. Jin, X. Wang, F. Cheng, S. Guo, Backdoor attacks on unsupervised graph representation learning, Neur. Networks, 180 (2024), 106668. https://doi.org/10.1016/j.neunet.2024.106668 doi: 10.1016/j.neunet.2024.106668
[44]	Y. Song, Research on sea surface temperature prediction based on spatiotemporal graph convolutional network, Master's Thesis, in Chinese, Nanjing University of Information Science and Technology, 2024.
[45]	W. Ren, Y. Li, M. Gao, X. Na, STCA-DeFormer: A spatio-temporal Transformer with seasonal decomposition and cross-modal attention for sea surface temperature prediction, Phys. Scr., 101 (2026), 046001. https://doi.org/1088/1402-4896/ae3706

Reader Comments

Your name:*

Email:*
© 2026 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)