Impact of physical and attention mechanisms on U-Net for SST forecasting

Abstract

Sea surface temperature (SST) prediction is a subject of great significance to the marine environment and human society. Changes in SST not only impact marine ecosystems and fishery resources but also trigger extreme weather events and disastrous consequences. Therefore, the precise prediction of SST is essential to avoiding these problems. Although numerous data-driven SST prediction models have emerged in recent years, these models are characterized by a lack of physical mechanisms related to sea temperature changes as well as insufficient generalization capabilities and interpretability. In our work, attempts were made to integrate physics-related convection phenomena into deep learning models, and traditional deep learning models were improved by incorporating time and space attention modules. The results of a series of experiments showed that the incorporation of physical mechanisms enhanced the performance of data-driven models. Furthermore, attention mechanisms were similarly helpful, of which temporal attention proved to be more important. The modules proposed in this work also improved the baseline model’s accuracy by 22%. In addition, seven-day SST predictions were carried out for the world’s five major fishing grounds. The results demonstrated that the application of transfer learning strategies yielded superior performance, further improving prediction accuracy by 1%–5%.

Keywords

SST prediction / U-Net / Attention mechanisms / Deep learning / Data-driven models

References

[1] Ali A, Zhu Y, Zakarya M. Exploiting dynamic spatio-temporal correlations for citywide traffic flow prediction using attention based neural networks. Inf Sci, 2021, 577: 852-870,

[2] Castro SL, Wick GA, Steele M. Validation of satellite sea surface temperature analyses in the Beaufort Sea using UpTempO buoys. Remote Sens Environ, 2016, 187: 458-475,

[3] Chandrasekhar S. Stochastic problems in physics and astronomy. Rev Mod Phys, 1943, 15(1): 1,

[4] Cheung WW, Frölicher TL, Lam VW, Oyinlola MA, Reygondeau G, Sumaila UR, et al.. Marine high temperature extremes amplify the impacts of climate change on fish and fisheries. Sci Adv, 2021, 7(40): eabh0895,

[5] Czaja A, Frankignoul C. Influence of the North Atlantic SST on the atmospheric circulation. Geophys Res Lett, 1999, 26(19): 2969-2972,

[6] Czaja A, Frankignoul C. Observed impact of Atlantic SST anomalies on the north atlantic Oscillation. J Clim, 2002, 15(6): 606-623,

[7] Danilov S, Kivman G, Schröter J. A finite-element ocean model: principles and evaluation. Ocean Model, 2004, 6(2): 125-150,

[8] Finn C, Goodfellow I, Levine S (2016) Unsupervised learning for physical interaction through video prediction. In: Proceedings of the 30th International Conference on Neural Information Processing Systems (NIPS 2016), Barcelona, pp 64–72

[9] Hollowed AB, Barange M, Beamish RJ, Brander K, Cochrane K, Drinkwater K, et al.. Projected impacts of climate change on marine fish and fisheries. ICES J Mar Sci, 2013, 70(5): 1023-1037,

[10] Kim D-H, Moon I-J, Lim C, Woo S-B. Improved prediction of extreme ENSO events using an artificial neural network with weighted loss functions. Front Mar Sci, 2023, 10: 1309609,

[11] Klein ES, Hill SL, Hinke JT, Phillips T, Watters GM. Impacts of rising sea temperature on krill increase risks for predators in the scotia sea. PLoS One, 2018, 13(1): e0191011,

[12] Koenigstein S, Mark FC, Gößling-Reisemann S, Reuter H, Poertner HO. Modelling climate change impacts on marine fish populations: process-based integration of ocean warming, acidification and other environmental drivers. Fish Fish, 2016, 17(4): 972-1004,

[13] Lau N-C. Interactions between global SST anomalies and the midlatitude atmospheric circulation. Bull Am Meteorol Soc, 1997, 78(1): 21-34,

[14] Liang Y, Li H, Guo B, Yu Z, Zheng X, Samtani S, et al.. Fusion of heterogeneous attention mechanisms in multi-view convolutional neural network for text classification. Inf Sci, 2021, 548: 295-312,

[15] Liu J, Zhang T, Han G, Gou Y. TD-LSTM: Temporal dependence-based LSTM networks for marine temperature prediction. Sensors, 2018, 18(11): 3797,

[16] McBride MM, Dalpadado P, Drinkwater KF, Godø OR, Hobday AJ, Hollowed AB, Kristiansen T, Murphy EJ, Ressler PH, Subbey S. Krill, climate, and contrasting future scenarios for arctic and antarctic fisheries. ICES J Mar Sci, 2014, 71(7): 1934-1955,

[17] Menter FR, Kuntz M, Langtry R (2003) Ten years of industrial experience with the SST turbulence model. In: Hanjalić K et al (eds) Turbulence, heat and mass transfer 4. Begell House Inc., pp 625–632

[18] Min X, Zhai G, Gu K, Liu J, Wang S, Zhang X, et al.. Visual attention analysis and prediction on human faces. Inf Sci, 2017, 420: 417-430,

[19] Peng S, Robinson WA, Li S. Mechanisms for the nao responses to the north atlantic SST tripole. J Clim, 2003, 16(12): 1987-2004,

[20] Raffel C, Shazeer N, Roberts A, Lee K, Narang S, Matena M, et al.. Exploring the limits of transfer learning with a unified text-to-text Transformer. J Mach Learn Res, 2020, 21(1): 140

[21] Reichstein M, Camps-Valls G, Stevens B, Jung M, Denzler J, Carvalhais N, et al.. Deep learning and process understanding for data-driven Earth system science. Nature, 2019, 566(7743): 195-204,

[22] Reynolds RW, Rayner NA, Smith TM, Stokes DC, Wang W. An improved in situ and satellite SST analysis for climate. J Clim, 2002, 15(13): 1609-1625,

[23] Sharma R, Basu S, Sarkar A, Pal P. Data-adaptive prediction of sea-surface temperature in the arabian sea. IEEE Geosci Remote Sens Lett, 2010, 8(1): 9-13

[24] Shaw P, Uszkoreit J, Vaswani A (2018) Self-attention with relative position representations. In: Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, New Orleans, pp 464–468

[25] Su H, Wu X, Yan XH, Kidwell A. Estimation of subsurface temperature anomaly in the indian ocean during recent global surface warming hiatus from satellite measurements: a support vector machine approach. Remote Sens Environ, 2015, 160: 63-71,

[26] Sun Y, Yao X, Bi X, Huang X, Zhao X, Qiao B. Time-series graph network for sea surface temperature prediction. Big Data Res, 2021, 25: 100237,

[27] Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN et al (2017) Attention is all you need. In: Proceedings of 31th International Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, pp 6000–6010

[28] Wang J, Chen X, Li Y, Boenish R. The effects of climate-induced environmental variability on Pacific ocean squids. ICES J Mar Sci, 2023, 80(4): 878-888,

[29] Wang Y, Zhang Y, Wang G-G. Forecasting ENSO using convolutional LSTM network with improved attention mechanism and models recombined by genetic algorithm in CMIP5/6. Inf Sci, 2023, 642: 119106,

[30] Weatherdon LV, Magnan AK, Rogers AD, Sumaila UR, Cheung WW. Observed and projected impacts of climate change on marine fisheries, aquaculture, coastal tourism, and human health: An update. Front Mar Sci, 2016, 3: 48,

[31] Wei L, Guan L, Qu L. Prediction of sea surface temperature in the South China Sea by artificial neural networks. IEEE Geosci Remote Sens Lett, 2019, 17(4): 558-562,

[32] Wenzel M, Zalesnyi VB. Data assimilation in a one-dimensional heat convection-diffusion model in the ocean. Izv Atmos Ocean Phys, 1996, 32: 564-579

[33] Wolff S, O'Donncha F, Chen B. Statistical and machine learning ensemble modelling to forecast sea surface temperature. J Mar Syst, 2020, 208: 103347,

[34] Xiao CJ, Chen NC, Hu CL, Wang K, Gong JY, Chen ZQ. Short and mid-term sea surface temperature prediction using time-series satellite data and LSTM-AdaBoost combination approach. Remote Sens Environ, 2019, 233: 111358,

[35] Xie J, Zhang J, Yu J, Xu L. An adaptive scale sea surface temperature predicting method based on deep learning with attention mechanism. IEEE Geosci Remote Sens Lett, 2019, 17(5): 740-744,

[36] Yang Y, Dong J, Sun X, Lima E, Mu Q, Wang X. A CFCC-LSTM model for sea surface temperature prediction. IEEE Geosci Remote Sens Lett, 2017, 15(2): 207-211,

[37] Yin Y, Le Guen V, Dona J, de Bézenac E, Ayed I, Thome N, et al.. (2021) Augmenting physical models with deep networks for complex dynamics forecasting. J Stat Mech-Theory Exp, 2021, 12: 124012,

[38] Yuan W, Wang H, Yu X, Liu N, Li Z. Attention-based context-aware sequential recommendation model. Inf Sci, 2020, 510: 122-134,

[39] Zhang Q, Wang H, Dong J, Zhong G, Sun X. Prediction of sea surface temperature using Long Short-Term Memory. IEEE Geosci Remote Sens Lett, 2017, 14(10): 1745-1749,