摘要
采用2010-2017年南海5个浮标波高观测资料和中国气象局热带气旋最佳路径集中的热带气旋参数,基于前馈型误差反向传播(Forward Feedback Back Propagation, FFBP)神经网络(Artificial Neural Network, ANN)方法,分别建立了各浮标站的台风浪高快速计算模型。研究显示,基于热带气旋中心坐标、中心最低气压、近中心最大风速、热带气旋中心与浮标之间的距离和方位4个参数建立的神经网络模型经反复训练后,模型输出结果可以很好地拟合观测数据,各浮标有效波高计算值与观测值的均方根误差小于0.3m,平均相对误差为5.78%~7.23%,相关系数大于0.9,属高度相关。独立测试结果显示,"山竹"(国际编号:1822)影响期间有效波高最大值的神经网络模型预报结果与观测值基本吻合,相对误差为–31.06%~0.98%,但计算的最大值出现时间和观测情况不完全一致。该计算方法可应用于热带气旋影响期间的有效波高最大值计算,因而在海洋工程领域和海洋预报领域具有应用前景。
Based on the wave height observation data of five buoys and parameters from the CMA Tropical Cyclone Database for the South China Sea from 2010 to 2017, fast calculation models of tropical cyclone-generated wave height are established by using Forward Feedback Back Propagation(FFBP) Artificial Neural Network(ANN). The results show that significant wave height calculated by the ANN model based on the tropical cyclone central location, minimum sea level pressure(MSLP), maximum sustained wind(MSW), the distance and azimuth between tropical cyclone center and buoy can well fit the observations. The root mean square errors between the calculation and observation of significant wave height are less than 0.3 m, the average relative errors are between 5.78% and 7.23%, and the correlation coefficients are greater than 0.9. Test results show that the model calculation of the maxima of significant wave height basically coincides with the observation, the relative errors being between –31.06% and 0.98%;however, when the calculated maxima of significant wave height appear is not in agreement with observation during tropical cyclone Mangkhut(International number ID: 1822). The method introduced in this paper can be applied to calculating the maximum of significant wave height during tropical cyclone for ocean engineering and marine forecast.
作者
周水华
洪晓
梁昌霞
江丽芳
ZHOU Shuihua;Hong Xiao;LIANG Changxia;JIANG Lifang(South China Sea Marine Forecast Center of State Oceanic Administration,Guangzhou 510310,China)
出处
《热带海洋学报》
CAS
CSCD
北大核心
2020年第4期25-33,共9页
Journal of Tropical Oceanography
基金
国家重点研发计划项目(2017YFC1404700)。
关键词
台风浪高
人工神经网络模型
权值
热带气旋参数
快速计算
tropical cyclone wave height
Artificial Neural Network Model
weight
tropical cyclone parameters
fast calculation
