| 摘要: |
| 内孤立波是发生在密度稳定层化海水中的一种特殊的海洋内波。预测内孤立波传播难度较大。本文提出了一种方法，利用美国麻省理工学院大气环流模型（MITgcm）的内孤立波模型计算了大量模拟数据，建立数据库。采用机器学习的方法，建立一个基于支持向量机（support vector machine，SVM）的安达曼海南部内孤立波传播预测模型。最后运用安达曼海南部的Sentinel-1A合成孔径雷达（SAR）图像对内孤立波传播预测模型结果进行检验。结果表明：基于SVM的内孤立波传播时间预测模型预测的时间平均绝对百分比误差为8.43%，平均绝对误差为1.00 h。基于SVM的内孤立波到达位置预测模型预测的位置平均绝对百分比误差为0.071%，平均绝对误差为0.069°。基于SVM的内孤立波振幅预测模型预测的振幅范围为23.80~84.98 m。 |
| 关键词: 内孤立波 MITgcm数值模拟 支持向量机(SVM) 预测 |
| DOI：10.11759/hykx20201105007 |
| 分类号:TP79 |
| 基金项目:国家重点研发项目（2017YFC1405600）；国家自然科学基金项目（61871353） |
|
| The prediction of internal solitary wave propagation characteristics in the southern Andaman Sea based on a support vector machine |
|
LU Ke-xiao, WANG Jing, WEI Xin
|
|
College of Physics and Optoelectronic Engineering, Ocean University of China, Qingdao 266100, China
|
| Abstract: |
| Internal solitary wave (ISW) is a special type of internal wave. It occurs in density-stable stratified seawater and have strong randomness. Their occurrence location, arrival time, amplitude, and other parameters are greatly affected by hydrology and other aspects of the external environment. Therefore, accurate prediction of internal solitary waves propagation is a recognized challenge. In this paper, a method is proposed to use the Massachusetts Institute of Technology general circulation model (MITgcm) of ISWs to calculate abundant simulation data and establish a database. A prediction model of ISWs propagation in the southern Andaman Sea based on a support vector machine (SVM) is established using machine learning. Finally, the Sentinel-1 synthetic aperture radar (SAR) images in the southern Andaman Sea are used to test the results of the ISWs propagation prediction model. The results demonstrate that the mean absolute percentage error of time (location) predicted by the prediction model of ISWs propagation time (arrival location) based on an SVM is 8.43% (0.072%), and the mean absolute error is 1.00 h (0.069°). The predicted amplitude range of the ISWs amplitude prediction model based on an SVM is 23.80~84.98 m. |
| Key words: internal solitary waves MITgcm numerical simulation support vector machine(SVM) prediction |
