融合IVMD的海表温度时空智能预测方法  

SST spatio-temporal intelligent prediction method integrating IVMD

作　　者：韩莹 曹允重[2] 张凌珺 赵芮晗 董昌明 HAN Ying;CAO Yunzhong;ZHANG Lingjun;ZHAO Ruihan;DONG Changming(Jiangsu atmospheric environment and equipment technology collaborative innovation center,Nanjing 210044,China;School of Automation,Nanjing University of Information Science and Technology,Nanjing 210044,China;Southern Marine Science and Engineering Guangdong Laboratory(Zhuhai),Zhuhai 519000,China;School of Ocean Science,Nanjing University of Information Science and Technology,Nanjing 210044,China)

机构地区：[1]江苏省大气环境与装备技术协同创新中心,江苏南京210044 [2]南京信息工程大学自动化学院,江苏南京210044 [3]南方海洋科学与工程广东省实验室(珠海),广东珠海519000 [4]南京信息工程大学海洋科学学院,江苏南京210044

出　　处：《海洋测绘》2024年第3期53-57,61,共6页Hydrographic Surveying and Charting

基　　金：国家自然科学基金(62076136);南方海洋科学与工程广东省实验室(珠海)基金(SML2020SP007)。

摘　　要：精准的海洋表面温度(sea surface temperature, SST)预测在海洋和气象领域具有重要意义,如海洋渔业捕捞和海洋天气预报等。提出一种融合改进变分模态分解(improved variational mode decomposition, IVMD)的时空混合模型来预测SST,采用中心频率观察法、残差指数最小化和皮尔逊相关系数改进变分模态分解(variational mode decomposition, VMD),去除SST序列冗余,利用图卷积神经网络(graph convolutional network, GCN)提取SST交互特征并结合长短时记忆网络(long short-term memory, LSTM)捕捉时间动态,提高预测精度。选取中国东海海域进行实证分析,实验结果表明:与现有模型对比,本文模型在均方根误差、平均绝对误差和平均绝对百分比误差3个指标上均有显著提升,验证了本文模型的有效性和稳定性。Accurate Sea Surface Temperature(SST)prediction is vital in marine and meteorological fields,such as marine fisheries and marine weather forecasting.A spatio-temporal hybrid model based on Improved Variational Mode Decomposition(IVMD)is proposed to predict SST.The Variational Mode Decomposition(VMD)method was improved by central frequency observation,residual index minimization and Pearson correlation coefficient to remove SST sequence redundancy.The Graph Convolutional Network(GCN)was adopted to extract SST interaction features,and Long Short-Term Memory(LSTM)was introduction to capture time dynamics.Combination of the above two model can enhance prediction accuracy.The East China Sea was selected for empirical analysis.Experimental results show that,compared with the existing model,the proposed model has significantly improved the root mean square error,mean absolute error and mean absolute percentage error.The effectiveness and stability of the proposed model are verified.

关 键 词：海洋表面温度预测 改进变分模态分解 皮尔逊相关系数 图卷积神经网络 长短时记忆网络 

分 类 号：P229.7[天文地球—大地测量学与测量工程]