|
摘要：
|
| 基于遮浪海洋站实测海温数据和气象数值预报数据，对传统的卷积神经网络（CNN）进行优化，采用优化CNN方法搭建近岸定点短期海温逐时预报模型，并开展建模研究。研究显示：将观测数据与气象数值预报数据相结合用于训练神经网络模型，能够有效提升模型海温预测效果；在CNN参数调节中选择合适的卷积核数量，模型可发挥出最优的海温预报性能；相比传统CNN、长短期记忆网络（LSTM）和持续性预报方法，优化CNN方法新增归一化层，模型在训练过程中能够更快更好地收敛，在测试数据集中具备更优的海温预测效果，在0～24 h、24～48 h、48～72 h预报时效内，该模型预报结果的平均绝对误差分别为0.276℃、0.412℃、0.487℃，均方根误差分别为0.459℃、0.632℃、0.714℃，相关系数分别为0.994 7、0.990 0、0.987 2。 |
| Based on the sea surface temperature(SST)observational data of Zhelang Station and the Numerical Weather Prediction(NWP) results, this study optimizes the traditional Convolutional Neural Networks(CNN) to construct a short-term forecasting model for nearshore hourly SST forecasts. The experimental results show that using the SST observational data and the NWP results to train the neural network model improves the SST prediction performance. The model performs best when the number of convolution kernel is tuned. Compared with the traditional CNN, LSTM model and persistence forecasting method, the optimized CNN model has better SST prediction in the testing set and faster learning speed through involving the normalized layer. The mean absolute errors of 0～24 h, 24～48 h, and 48～72 h forecasts are 0.276 ℃, 0.412 ℃ and 0.487 ℃, the corresponding root mean square errors are 0.217 ℃, 0.396 ℃ and 0.567 ℃, and the corresponding correlation coefficients are 0.994 7, 0.990 0, and 0.987 2, respectively. |
|
参考文献：
|
[1] 刘娜,王辉,凌铁军,等.全球业务化海洋预报进展与展望[J].地球科学进展, 2018, 33(2):131-140.Liu Na, Wang Hui, Ling Tiejun, et al.Review and prospect of global operational ocean forecasting[J].Advances in Earth Science,2018, 33(2):131-140.
[2] 张永垂,陈诗尧,王宁,等.全球业务化海洋预报系统进展[J].地球科学进展, 2022, 37(4):344-357.Zhang Yongchui, Chen Shiyao, Wang Ning, et al.Progress of global operational ocean forecasting systems[J].Advances in Earth Science, 2022, 37(4):344-357.
[3] 张建华.海温预报知识讲座第一讲海水温度预报概况[J].海洋预报, 2003, 20(4):81-85.Zhang Jianhua.Lecture on sea water temperature prediction:Lecture 1 overview of sea water temperature prediction[J].Marine Forecasts, 2003, 20(4):81-85.
[4] 王兆毅,李云,王旭.中国近岸海域基础预报单元海温预报指导产品研制[J].海洋预报, 2020, 37(4):59-65.Wang Zhaoyi, Li Yun, Wang Xu.Development of forecast guidance product for sea temperature of basic forecast units in the Chinese coastal waters[J].Marine Forecasts, 2020, 37(4):59-65.
[5] 张建华.海温预报知识讲座第二讲数理统计方法在海温预报中的应用[J].海洋预报, 2004, 21(1):85-90.Zhang Jianhua.Lecture on sea water temperature prediction:Lecture 2 application of mathematical statistical methods on sea water temperature prediction[J].Marine Forecasts, 2004, 21(1):85-90.
[6] 赵强,王擎宇,舒志光.基于SARIMA模型的近岸海表温度短期预报研究[J].海洋预报, 2024, 41(1):42-49.Zhao Qiang, Wang Qingyu, Shu Zhiguang.A study of short-term forecast of nearshore SST based on SARIMA model[J].Marine Forecasts, 2024, 41(1):42-49.
[7] Zhang Qin, Wang Hui, Dong Junyu, et al.Prediction of sea surface temperature using long short-term memory[J].IEEE Geoscience and Remote Sensing Letters, 2017, 14(10):1745-1749.
[8] Zhang Xiaoyu, Li Yongqing, Frery A C, et al.Sea surface temperature prediction with memory graph convolutional networks[J].IEEE Geoscience and Remote Sensing Letters, 2022, 19:8017105.
[9] 贺琪,查铖,宋巍,等.基于STL的海表面温度预测算法[J].海洋环境科学, 2020, 39(6):918-925.He Qi, Zha Cheng, Song Wei, et al.Sea surface temperature prediction algorithm based on STL model[J].Marine Environmental Science, 2020, 39(6):918-925.
[10] 杨乐晴,王丽娜,张红春,等.基于STL的南海海表温度组合预测模型[J].海洋环境科学, 2024, 43(1):109-118.Yang Leqing, Wang Lina, Zhang Hongchun, et al.Combining forecasting model for sea surface temperature in the South China Sea based on STL[J].Marine Environmental Science, 2024, 43(1):109-118.
[11] 匡晓迪,王兆毅,张苗茵,等.基于BP神经网络方法的近岸数值海温预报释用技术[J].海洋与湖沼, 2016, 47(6):1107-1115.Kuang Xiaodi, Wang Zhaoyi, Zhang Miaoyin, et al.An interpretation scheme of numerical near-shore sea-water temperature forecast based on BPNN[J].Oceanologia et Limnologia Sinica, 2016, 47(6):1107-1115.
[12] 林小刚,王兆毅,李竞时,等.基于LSTM神经网络方法的粤东近岸海温预报[J].海洋预报, 2022, 39(5):27-36.Lin Xiaogang, Wang Zhaoyi, Li Jingshi, et al.Sea temperature forecasting based on LSTM neural network along the coast of eastern Guangdong[J].Marine Forecasts, 2022, 39(5):27-36.
[13] 李竞时,匡晓迪,李琼,等.基于主成分分析和LSTM神经网络的海温预报模型[J].海洋预报, 2023, 40(2):1-10.Li Jingshi, Kuang Xiaodi, Li Qiong, et al.SST forecasting model based on principal component analysis and LSTM neural network[J].Marine Forecasts, 2023, 40(2):1-10.
[14] 何恩业,李琼,张聿柏,等.基于PCA-BP特征工程的近海单点海温预报技术及应用[J].海洋预报, 2023, 40(3):35-44.He Enye, Li Qiong, Zhang Yubai, et al.Technology and application of offshore SST prediction based on PCA-BP feature engineering[J].Marine Forecasts, 2023, 40(3):35-44.
[15] 赖晓倩,余镒琦,梁中耀,等.基于差分回归模型和可迁移长短期记忆网络集成的三沙湾水温预测[J].海洋学报, 2023, 45(4):165-178.Lai Xiaoqian, Yu Yiqi, Liang Zhongyao, et al.Water temperature prediction in the Sansha Bay based on the integration of differential regression model and transportable long short-term memory network[J].Haiyang Xuebao, 2023, 45(4):165-178.
[16] Jiao Yan, Li Ge, Zhao Peng, et al.Construction of sea surface temperature forecasting model for Bohai Sea and Yellow Sea coastal stations based on long short-time memory neural network[J].Water, 2024, 16(16):2307.
[17] 李双林,张仲石,王惠.数值天气预报的未来是人工智能与数学物理模型的融合?[J].地球科学, 2022, 47(10):3919-3921.Li Shuanglin, Zhang Zhongshi, Wang Hui.Will the future of numerical weather prediction be a fusion of artificial intelligence and mathematical and physical modeling?[J].Earth Science,2022, 47(10):3919-3921.
[18] Van Den Oord A, Dieleman S, Zen Heiga, et al.WaveNet:A generative model for raw audio[C] //Proceedings of the 9th ISCA Workshop on Speech Synthesis(SSW 9).Sunnyvale:ISCA, 2016.
[19] Borovykh A, Bohte S, Oosterlee C W.Conditional time series forecasting with convolutional neural networks[PP/OL].V2.arXiv(2017-05-16)[2024-09-29].https://arxiv.org/abs/1703.04691v2.
[20] Bai Shaojie, Kolter J Z, Koltun V.An empirical evaluation of generic convolutional and recurrent networks for sequence modeling[PP/OL].V2.arXiv(2018-4-19)[2024-09-29].https://doi.org/10.48550/arXiv.1803.01271.
[21] 翁少佳,蔡锦海,庞运禧,等.卷积神经网络在近岸表层海温预报中的应用[J].热带海洋学报, 2024, 43(1):40-47.Weng Shaojia, Cai Jinhai, Pang Yunxi, et al.Application of convolutional neural network to sea surface temperature prediction in the coastal waters[J].Journal of Tropical Oceanography, 2024, 43(1):40-47.
[22] Ioffe S, Szegedy C.Batch normalization:Accelerating deep network training by reducing internal covariate shift[C] //Proceedings of the 32nd International Conference on Machine Learning.Online:PMLR, 2015:448-456.
[23] 张云翼,江毓武.汕尾外侧冷水跨陆架输送的形成机制[J].厦门大学学报(自然科学版), 2012, 51(4):746-752.Zhang Yunyi, Jiang Yuwu.The mechanism of cold water crossshelf transport in the continental shelf off Shanwei[J].Journal of Xiamen University(Natural Science), 2012, 51(4):746-752. |
|
服务与反馈：
|
|
【文章下载】【发表评论】【查看评论】【加入收藏】
|
|
|
