基于ConvLSTM的中国东南沿海波浪智能预报和评估

金阳; 韩磊; 金梅兵; 董昌明

doi:10.3969/j.issn.1001-909X.2024.03.007

PDF(4833 KB)

海洋学研究 ›› 2024, Vol. 42 ›› Issue (3) : 88-98. DOI: 10.3969/j.issn.1001-909X.2024.03.007

研究论文

基于ConvLSTM的中国东南沿海波浪智能预报和评估

金阳 ¹ ,
韩磊 ¹ ,
金梅兵 ¹^,²^,^* ,
董昌明 ¹^,²

作者信息 +

Intelligent wave forecasting and evaluation along the southeast coast of China based on ConvLSTM method

JIN Yang ¹ ,
HAN Lei ¹ ,
JIN Meibing ¹^,²^,^* ,
DONG Changming ¹^,²

Author information +

文章历史 +

摘要

相较于半理论半分析和数值模型的波浪预报方法,智能波浪预报有着精度高、计算资源需求低的优势。该文基于卷积长短期记忆网络(convolutional long short-term memory network, ConvLSTM)算法,建立了有效波高(significant wave height, SWH)二维预报模型,以中国东南沿海2014—2022年ERA5数据进行训练,通过敏感性试验优化模型配置,并开展中国东南沿海SWH在2023年4个预报时效(6 h、12 h、18 h、24 h)下的预测性能评估。敏感性试验显示,输入时间序列长度N=4(即输入-18 h, -12 h, -6 h, 0 h的SWH值)时,模型在4个预报时效下的准确性均优于其他时间序列长度;输入物理要素组合为SWH、平均波向和海面10 m 风矢量时,模型在12 h、18 h和24 h预报时效下的准确性优于其他组合。通过对ConvLSTM模型训练及配置的精细调整,可以实现对中国东南沿海SWH的二维、高精度的智能预报。

Abstract

Compared with the semi-theoretical and semi-analytical wave forecasting and numerical modeling,artificial intelligence wave forecasting has the advantages of higher forecasting accuracy and lower computational resource requirements. In this paper, a two-dimensional significant wave height (SWH) forecasting model for the southeast coast of China is established based on the convolutional long short-term memory network (ConvLSTM) algorithm using ERA5 (ECMWF Reanalysis v5) reanalysis data as the initial field. The data from 2014-2022 are used to train the forecasts of SWH for the next 6 h, 12 h, 18 h and 24 h, and the data from 2023 are used for testing. Sensitivity tests are carried out to optimize the model configuration and evaluate the prediction performance of SWH in the southeast coast of China at four period validity (6 h, 12 h, 18 h, 24 h) in 2023. Sensitivity tests show that when input time series length N=4(input SWH value of -18 h, -12 h, -6 h, 0 h), the accuracies of the model at four period validity are better than those of other time series length. When the combination of input physical elements is SWH, mean wave direction and sea surface 10 m wind vector, the accuracy of the model is better than other combinations at 12 h, 18 h and 24 h. Through the fine-tuning of ConvLSTM model training and configuration, the two-dimensional and high-precision intelligent prediction of SWH in the southeast coast of China can be realized.

导出引用

金阳, 韩磊, 金梅兵, 等. 基于ConvLSTM的中国东南沿海波浪智能预报和评估[J]. 海洋学研究. 2024, 42(3): 88-98 https://doi.org/10.3969/j.issn.1001-909X.2024.03.007

JIN Yang, HAN Lei, JIN Meibing, et al. Intelligent wave forecasting and evaluation along the southeast coast of China based on ConvLSTM method[J]. Journal of Marine Sciences. 2024, 42(3): 88-98 https://doi.org/10.3969/j.issn.1001-909X.2024.03.007

中图分类号： P731.33;P714

参考文献

列表( 原文顺序 | 文献年度倒序 | 文中引用次数倒序 ) 可视化分析

[1]	SHENG W N. Wave energy conversion and hydrodynamics modelling technologies: A review[J]. Renewable and Sustainable Energy Reviews, 2019, 109: 482-498. 本文引用 [1]

[2]	REGUERO B G, LOSADA I J, MÉNDEZ F J. A recent increase in global wave power as a consequence of oceanic-warming[J]. Nature Communications, 2019, 10: 205. 本文引用 [1]

[3]	KAMPHUIS J W. Introduction to coastal engineering and management[M]. Singapore: World Scientific, 2020. 本文引用 [1]

[4]	WANG W X, TANG R C, LI C, et al. A BP neural network model optimized by mind evolutionary algorithm for predicting the ocean wave heights[J]. Ocean Engineering, 2018, 162: 98-107. 本文引用 [2]

[5]	YEGANEH-BAKHTIARY A, EYVAZOGHLI H, SHABAKHTY N, et al. Machine learning prediction of wave characteristics: Comparison between semi-empirical approaches and DT model[J]. Ocean Engineering, 2023, 286: 115583. 本文引用 [1]

[6]	刘凡, 陆小敏, 徐丹, 等. 海浪预报方法研究进展[J]. 河海大学学报:自然科学版, 2021, 49(5):387-393. LIU F, LU X M, XU D, et al. Research progress of ocean waves forecasting method[J]. Journal of Hohai University: Natural Sciences, 2021, 49(5): 387-393. 本文引用 [1]

[7]	BRETSCHNEIDER C L. Hurricane design wave practices[J]. Journal of the Waterways and Harbors Division, 1957, 83: 1-33. 本文引用 [1]

[8]	NEUMANN G, PIERSON W J. A detailed comparison of theoretical wave spectra and wave forecasting methods[J]. Deutsche Hydrographische Zeitschrift, 1957, 10(4): 134-146. 本文引用 [1]

[9]	文圣常. 普遍风浪谱及其应用[J]. 山东海洋学院学报,1960:15-43. WEN S C. Universal wind wave spectrum and its applications[J]. Journal of Shandong Ocean University, 1960: 15-43. 本文引用 [1]

[10]	KAMRANZAD B, ETEMAD-SHAHIDI A, KAZEMINEZHAD M H. Wave height forecasting in Dayyer, the Persian Gulf[J]. Ocean Engineering, 2011, 38(1): 248-255. 本文引用 [1]

[11]	GELCI R, CAZALÉ H, VASSAL J. La méthode des densités spectroangulaires[J]. Comité Central Océanogr, 1957, 9: 416-435. 本文引用 [1]

[12]	DURRANT T, GREENSLADE D, HEMER M, et al. A global wave hindcast focussed on the Central and South Pacific[R]//The Centre for Australian Weather and Climat Research Technical Report No. 070, 2014. 本文引用 [1]

[13]	BOOIJ N, RIS R C, HOLTHUIJSEN L H. A third-generation wave model for coastal regions: 1. Model descrip-tion and validation[J]. Journal of Geophysical Research: Oceans, 1999, 104(C4): 7649-7666. 本文引用 [1]

[14]	TOLMAN H L. User manual and system documentation of WAVEWATCH-III TM version 3.14[R]. [S.l.]: NECP, 2009. 本文引用 [1]

[15]	文圣常, 钱成春, 叶安乐, 等. 基于选定风浪方向谱的海浪模拟方法(英文)[J]. 青岛海洋大学学报, 1999(3): 345-397. WEN S C, QIAN C C, YE A L, et al. Wave modeling based on adopted wind-wave direction spectrum[J]. Journal of Ocean University of Qingdao, 1999 (3): 345-397. 本文引用 [1]

[16]	袁业立, 华锋, 潘增弟, 等. LAGFD-WAM海浪数值模式:Ⅱ.区域性特征线嵌入格式及其应用[J]. 海洋学报, 1992, 14(6):12-24. YUAN Y L, HUA F, PAN Z D, et al. LAGFD-WAM numerical model of ocean waves—Ⅱ. Embedding scheme of regional characteristic lines and its application[J]. Acta Oceanologica Sinica, 1992, 14(6): 12-24. 本文引用 [1]

[17]	MAKARYNSKYY O. Improving wave predictions with artificial neural networks[J]. Ocean Engineering, 2004, 31(5/6): 709-724. 本文引用 [2]

[18]	ZUBIER K. Using an artificial neural network for wave height forecasting in the Red Sea[J]. Indian Journal of Geo Marine Sciences, 2020, 49: 184-191. 本文引用 [1]

[19]	CORNEJO-BUENO L, NIETO-BORGE J C, GARCÍA-DÍAZ P, et al. Significant wave height and energy flux prediction for marine energy applications: A grouping genetic algorithm-extreme learning machine approach[J]. Renewable Energy, 2016, 97: 380-389. 本文引用 [1]

[20]	KUMAR N K, SAVITHA R, MAMUN A A. Ocean wave height prediction using ensemble of extreme learning machine[J]. Neurocomputing, 2018, 277: 12-20. 本文引用 [1]

[21]	PUSHPAM P M M, ENIGO V S F. Forecasting significant wave height using RNN-LSTM Models[C]//2020 4th Inter-national Conference on Intelligent Computing and Control Systems (ICICCS). Madurai, India. IEEE, 2020: 1141-1146. 本文引用 [2]

[22]	GERS F A, SCHRAUDOLPH N N, SCHMIDHUBER J. Learning precise timing with LSTM recurrent networks[J]. Journal of Machine Learning Research, 2002, 3: 115-143. 本文引用 [1]

[23]

HOCHREITER

, SCHMIDHUBER

. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.

https://doi.org/10.1162/neco.1997.9.8.1735

https://www.ncbi.nlm.nih.gov/pubmed/9377276

本文引用 [1] 摘要

Learning to store information over extended time intervals by recurrent backpropagation takes a very long time, mostly because of insufficient, decaying error backflow. We briefly review Hochreiter's (1991) analysis of this problem, then address it by introducing a novel, efficient, gradient-based method called long short-term memory (LSTM). Truncating the gradient where this does not do harm, LSTM can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units. Multiplicative gate units learn to open and close access to the constant error flow. LSTM is local in space and time; its computational complexity per time step and weight is O(1). Our experiments with artificial data involve local, distributed, real-valued, and noisy pattern representations. In comparisons with real-time recurrent learning, back propagation through time, recurrent cascade correlation, Elman nets, and neural sequence chunking, LSTM leads to many more successful runs, and learns much faster. LSTM also solves complex, artificial long-time-lag tasks that have never been solved by previous recurrent network algorithms.

[24]	HU H G, VAN DER WESTHUYSEN A J, CHU P, et al. Predicting Lake Erie wave heights and periods using XGBoost and LSTM[J]. Ocean Modelling, 2021, 164: 101832. 本文引用 [1]

[25]	FAN S T, XIAO N H, DONG S. A novel model to predict significant wave height based on long short-term memory network[J]. Ocean Engineering, 2020, 205: 107298. 本文引用 [1]

[26]	SHI X J, CHEN Z R, WANG H, et al. Convolutional LSTM network: A machine learning approach for precipitation nowcasting[J]. Advances in Neural Information Processing Systems, 2015: 802-810. 本文引用 [2]

[27]	CHOI H, PARK M, SON G, et al. Real-time significant wave height estimation from raw ocean images based on 2D and 3D deep neural networks[J]. Ocean Engineering, 2020, 201: 107129. 本文引用 [1]

[28]	ZHOU S Y, XIE W H, LU Y X, et al. ConvLSTM-based wave forecasts in the South and East China Seas[J]. Frontiers in Marine Science, 2021, 8: 680079. 本文引用 [1]