基于自注意力机制的深度学习的海洋三维温度场预测-海洋预报

基于自注意力机制的深度学习的海洋三维温度场预测

单位：1. 青岛科技大学, 山东青岛 266061;
2. 中国科学院海洋研究所, 山东青岛 266071;
3. 中国科学院大学, 北京 100094;
4. 青岛海洋科技中心, 山东青岛 266000

关键词：海水温度三维温度场预测自注意力记忆机制 SA-ConvLSTM 多步长递归预测

分类号：P731.31

出版年·卷·期（页码）：2024·41·第三期（22-32）

摘要：

目前主要从时空角度出发对海洋三维温度场进行预测，却忽略了相邻位置的海温相关关系。为弥补这一不足，构建一种融合了自注意力记忆模块与卷积式长短时记忆神经网络（ConvLSTM）模型的 SA-ConvLSTM 三维温度场预测模型，不仅可以从历史三维温度场中提取海温时空特征，还能获取并记忆相邻点位置信息，从而实现对三维温度场时空变化的把握。实验结果表明：相较于ConvLSTM模型，SA-ConvLSTM模型在滑动预测与多步长递归预测实验下的均方根误差和平均绝对误差提升约 14%，且整体预测效果均优于基线模型、长短时记忆神经网络模型和ConvLSTM模型。

While previous researches on the 3-D ocean temperature field prediction mainly focused on the perspective of spatial and temporal relationship which ignored the relationship of relative location, this article proposes a SA-ConvLSTM 3-D ocean temperature field prediction model which combines self-attention memory module and ConvLSTM. The new model is not only able to extract the spatial-temporal features in historical 3-D ocean temperature fields, but can obtain and record the information of location to learn the laws of seawater in both space and time. The experimental results show that the RMSE and MAE of the SA-ConvLSTM forecasts have approximately increased by 14% in sliding prediction and multi-step recursive prediction, and its overall performance is better than the persistence, LSTM and ConvLSTM model. Our research provides a new idea for the prediction of the 3-D seawater temperature field.

参考文献：

[1] AWAL M A, KHAN M A H. Global warming and sea level rising:impact on agriculture and food security in southern coastal region of Bangladesh[J]. Asian Journal of Geographical Research, 2020, 3(3):9-36.
[2] SOLANKI H U, BHATPURIA D, CHAUHAN P. Integrative analysis of AltiKa-SSHa, MODIS-SST, and OCM-Chlorophyll signatures for fisheries applications[J]. Marine Geodesy, 2015, 38(S1):672-683.
[3] NEUMANN D R. Seasonal movements of short-beaked common dolphins (Delphinus delphis) in the north-western Bay of Plenty, New Zealand:Influence of sea surface temperature and El Niño/La Niña[J]. New Zealand Journal of Marine and Freshwater Research, 2001, 35(2):371-374.
[4] ROPELEWSKI C F, HALPERT M S. Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation[J]. Monthly Weather Review, 1987, 115(8):1606-1626.
[5] HAMMER G L, NICHOLLS N, MITCHELL C. Applications of seasonal climate forecasting in agricultural and natural ecosystems[M]. Dordrecht:Springer, 2000.
[6] BALMASEDA M A, TRENBERTH K E, KÄLLéN E. Distinctive climate signals in reanalysis of global ocean heat content[J]. Geophysical Research Letters, 2013, 40(9):1754-1759.
[7] CHENG L J, ZHU J. 2017 was the warmest year on record for the global ocean[J]. Advances in Atmospheric Sciences, 2018, 35(3):261-263.
[8] WANG Q, LIN J Z, YUAN Y. Salient band selection for hyperspectral image classification via manifold ranking[J]. IEEE Transactions on Neural Networks and Learning Systems, 2016, 27(6):1279-1289.
[9] STOCKDALE T N, BALMASEDA M A, VIDARD A. Tropical Atlantic SST prediction with coupled ocean-atmosphere GCMs[J]. Journal of Climate, 2006, 19(23):6047-6061.
[10] ZHANG Q, WANG H, DONG J Y, et al. Prediction of sea surface temperature using long short-term memory[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(10):1745-1749.
[11] YANG Y T, DONG J Y, SUN X, et al. A CFCC-LSTM model for sea surface temperature prediction[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(2):207-211.
[12] SHI X J, CHEN Z R, WANG H, et al. Convolutional LSTM network:A machine learning approach for precipitation nowcasting[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems. Montreal:MIT Press, 2015:802-810.
[13] ZHANG K, GENG X P, YAN X H. Prediction of 3-D ocean temperature by multilayer convolutional LSTM[J]. IEEE Geoscience and Remote Sensing Letters, 2020, 17(8):1303-1307.
[14] ZUO X Y, ZHOU X F, GUO D Q, et al. Ocean temperature prediction based on stereo spatial and temporal 4-D convolution model[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19:1-5.
[15] XIAO C J, TONG X H, LI D D, et al. Prediction of long lead monthly three-dimensional ocean temperature using time series gridded Argo data and a deep learning method[J]. International Journal of Applied Earth Observation and Geoinformation, 2022, 112:102971.
[16] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach:Curran Associates Inc., 2017:6000-6010.
[17] FU J, LIU J, TIAN H J, et al. Dual attention network for scene segmentation[C]//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach:IEEE, 2019:3141-3149.
[18] HUANG Z L, WANG X G, HUANG L C, et al. CCNet:Crisscross attention for semantic segmentation[C]//Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Seoul:IEEE, 2019:603-612.
[19] CHEN Y P, KALANTIDIS Y, LI J S, et al. A²-nets:Double attention networks[C]//Proceedings of the 32nd International Conference on Neural Information Processing Systems. Montréal:Curran Associates Inc., 2018:350-359.
[20] LIN Z H, LI M M, ZHENG Z B, et al. Self-attention ConvLSTM for spatiotemporal prediction[C]//Proceedings of the 34th AAAI Conference on Artificial Intelligence. New York:AAAI, 2020:11531-11538.
[21] ORSI A H, WHITWORTH III T, NOWLIN JR W D. On the meridional extent and fronts of the Antarctic Circumpolar Current[J]. Deep Sea Research Part I:Oceanographic Research Papers, 1995, 42(5):641-673.
[22] LI H, XU F H, ZHOU W, et al. Development of a global gridded Argo data set with Barnes successive corrections[J]. Journal of Geophysical Research:Oceans, 2017, 122(2):866-889.
[23] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas:IEEE, 2016:770-778.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】