[1]吴玉霜,黄小燕,陈家正,等.机器学习在广西台风极大风速预报中的应用[J].气象研究与应用,2021,42(04):26-31.[doi:10.19849/j.cnki.CN45-1356/P.2021.4.05]
　Wu Yushuang,Huang Xiaoyan,Chen Jiazheng,et al.Application of machine learning in forecasting maximum wind speed of typhoon in Guangxi[J].Journal of Meteorological Research and Application,2021,42(04):26-31.[doi:10.19849/j.cnki.CN45-1356/P.2021.4.05]
点击复制

机器学习在广西台风极大风速预报中的应用()
分享到：

气象研究与应用[ISSN:1673-8411/CN:45-1356/P]

卷:
第42卷
期数:
2021年04期
页码:
26-31
栏目:
广西台风团队专栏
出版日期:
2022-01-27

文章信息/Info

Title:
Application of machine learning in forecasting maximum wind speed of typhoon in Guangxi
作者:
吴玉霜1; 2;  黄小燕3;  陈家正4;  赵华生3
1. 广西壮族自治区气象台，南宁 530022;
2. 广西壮族自治区气象灾害防御技术中心，南宁 530022;
3. 广西壮族自治区气象科学研究所，南宁 530022;
4. 广西民族大学，南宁 530006
Author(s):
Wu Yushuang1; 2;  Huang Xiaoyan3;  Chen Jiazheng4;  Zhao Huasheng3
1. Guangxi Meteorological Observatory, Nanning Guangxi 530022, China;
2. Guangxi Meteorological Disaster Prevention Technology Center, Nanning Guangxi 530022, China;
3. Guangxi Institute of Meteorological Sciences, Nanning Guangxi 530022, China;
4. Guangxi University of Nationalities, Nanning Guangxi 530006, China
关键词:
机器学习; 台风; 极大风; 预报建模
Keywords:
machine learning; typhoon; maximum wind; forecast modeling
分类号:
P457.8
DOI:
10.19849/j.cnki.CN45-1356/P.2021.4.05
摘要:
以1980—2020年广西台风期间桂林、梧州、龙州、南宁、玉林等5个气象观测站的地面日极大风速为研究对象，采用多元线性回归（MR）、支持向量机（SVM）、模糊神经网络（FNN）等三种较为常用的线性和非线性方法分别进行预报建模，对2011—2020年共10a独立样本的检验。结果表明，在全样本风速预报的平均绝对误差上，FNN模型对桂林站、梧州站、龙州站、玉林站共4个站点预报的平均绝对误差最小，总体预报精度最好，MR预报模型则对南宁站有较好的预报能力，SVM模型预报效果总体偏差。对于6级以上大风的TS评分、命中率、空报率和预报偏差等4个评估指标的统计，FNN模型的预测精度最高且相对稳定，MR方案次之，SVM在三种方案中预报效果最差。FNN方法对广西台风期间地面日极大风速的预报有较好的参考作用。
Abstract:
Taking the ground daily maximum wind speeds of five meteorological observation stations in Guilin, Wuzhou, Longzhou, Nanning and Yulin during the typhoon in Guangxi from 1980 to 2020 as the research object, three commonly used linear and nonlinear methods such as multiple linear regression(MR), support vector machine(SVM), and fuzzy neural network(FNN)were used for prediction modeling respectively, and the independent samples from 2011 to 2020 were tested. The results show that in terms of the average absolute error of full-sample wind speed prediction, the average absolute error of FNN model for Guilin, Wuzhou, Longzhou and Yulin is the smallest, and the overall prediction accuracy is the best; MR prediction model has better prediction ability for Nanning station, while the prediction effect of SVM model is generally poor. The statistical results of the four evaluation indexes of TS score, hit rate, false alarm rate and prediction deviation for strong winds above level 6 show that the prediction accuracy of FNN model is the highest and relatively stable, followed by MR, and SVM is the worst among the three schemes. FNN method has a good reference for the prediction of daily maximum wind speed on the ground during typhoon in Guangxi.

参考文献/References:

[1] Knofczynski G T. MundfromD. Sample Sizes when Using Multiple Linear Regression for Prediction. [J]. Educational & Psychological Measurement, 2008, 68(3): 431-442.
[2] 修春波, 任晓, 李艳晴, 等. 基于卡尔曼滤波的风速序列短期预测方法[J]. 电工技术学报, 2014, 29(2): 253-259.
[3] Nayak M A, Ghosh S. Prediction of Extreme Rainfall Event Using Weather Pattern Recognition and Support Vector Machine Classifier[J]. Theoretical & Applied Climatology, 2013, 114(3-4): 583-603.
[4] Cevik H H, Cunkas M, Polat K. A New Multistage Short-term Wind Power Forecast Model using Decomposition and Artificial Intelligence Methods[J]. Physica A: Statistical Mechanics and its Applications, 2019(534): 122-177.
[5] 勾海芝, 赵征, 夏子涵. 基于经验模式分解的神经网络组合风速预测研究[J]. 电力科学与工程, 2017, 33(10): 62-67.
[6] 杨晓君, 张楠, 陈宏, 等. 基于人工神经网络算法的渤海海风预报方法研究[J]. 干旱气象, 2019, 37(1): 146-152.
[7] 朱智慧, 黄宁立, 王延凤. 台风影响期间上海沿海极大风速预报方法研究[A]. 中国气象学会. 创新驱动发展提高气象灾害防御能力—S2灾害天气监测、分析与预报[C]. 中国气象学会, 2013: 8.
[8] 钱燕珍, 孙军波, 余晖, 等. 用支持向量机方法做登陆热带气旋站点大风预报[J]. 气象, 2012, 38(3): 300-306.
[9] 肖志祥, 姚才, 赵金彪, 等. 广西热带气旋预报业务和研究进展[J]. 气象研究与应用, 2020, 41(4): 20-27.
[10] 金龙, 黄颖, 姚才, 等. 人工智能技术的热带气旋预报综述(之一)——BP神经网络和集成方法的热带气旋预报研究和业务应用[J]. 气象研究与应用, 2020, 41(2): 1-6.
[11] 郭彬, 卓健, 周冬静, 等. 基于智能计算的广西大风短临预报预警系统的产品检验[J]. 气象研究与应用, 2021, 42(1): 80-84.
[12] 陈润珍, 孔宁谦. 广西沿海热带气旋大风数值预报探讨[J]. 海洋预报, 2004(1): 52-55.
[13] 董彦, 林开平, 黄小燕. 南海热带气旋大风的遗传-神经网络集合预报[J]. 气象研究与应用, 2014, 35(1): 40-45.
[14] 李忠, 刘景霞. 基于遗传算法和最小二乘支持向量机的风电场超短期风速预测[J]. 电工技术, 2021(13): 56-59+66.
[15] 钱斌凯, 何彩芬, 金炜, 等. 基于支持向量机的多因子风速预测[J]. 宁波大学学报(理工版), 2018, 31(3): 14-19.
[16] 黄小燕, 史旭明, 刘苏东, 等. 模糊神经网络方法在热带气旋强度预报中的应用研究[J]. 高原气象, 2009, 28(6): 1408-1413.

相似文献/References:

[1]赖珍权,翟丽萍,古文保.1415号台风“海鸥”的卫星云图及雷达资料分析[J].气象研究与应用,2017,38(02):10.
　Lai Zhenquan,Zhai Liping,Gu Wenbao.Analysis of typhoon Kalmaegi(1415) based on satellite images and radar data[J].Journal of Meteorological Research and Application,2017,38(04):10.
[2]麦雪湖,植江玲,麦文强.台风“彩虹”(1522)的数值模拟及增强条件分析[J].气象研究与应用,2017,38(01):30.
　Mai Xue-hu,Zhi Jiang-ling,Mai Wen-qiang.Analysis on the numerical simulation and enhancing conditions of typhoon “Mujigea”[J].Journal of Meteorological Research and Application,2017,38(04):30.
[3]黄卓,廖雪萍.2016年台风“莎莉嘉”对广西的影响评估[J].气象研究与应用,2017,38(01):40.
　Huang Zhuo,Liao Xue-ping.Impact assessment of typhoon “Sarika” on Guangxi in 2016[J].Journal of Meteorological Research and Application,2017,38(04):40.
[4]李妍君,陈耀登.台风“凡亚比”的诊断分析和数值模拟研究[J].气象研究与应用,2017,38(03):1.
　Li Yan-jun,Chen Yao-deng.Typhoon “Fanapi” diagnostic analysis and numerical simulation research[J].Journal of Meteorological Research and Application,2017,38(04):1.
[5]钟利华,李仲怡,李勇,等.西江流域台风暴雨面雨量分布特征及天气概念模型[J].气象研究与应用,2017,38(03):13.
　Zhong Li-hua,Li Zhong-yi,Li Yong,et al.Areal rainfall distribution characteristics of typhoon torrential rain and weather concept model of Xijiang basin[J].Journal of Meteorological Research and Application,2017,38(04):13.
[6]郑志阳,杨苏勤,刘德和,等.影响福建省惠安县台风气候特征及其不同路径的风雨影响[J].气象研究与应用,2016,37(01):34.
　Zheng Zhiyang,Yang Suqin,Liu Dehe,et al.Influence of fujian province huian typhoon climate characteristics and the effect of different paths[J].Journal of Meteorological Research and Application,2016,37(04):34.
[7]刘远方.台风“海鸥”影响南宁机场强降水过程分析[J].气象研究与应用,2016,37(01):56.
　Liu yuan-fang.The analysis of heavy rainfall process during Nanning airport affected by typhoon “kalmaegi”[J].Journal of Meteorological Research and Application,2016,37(04):56.
[8]周渭,蒋平凡,黎馨.2015台风“彩虹”强降雨过程分析[J].气象研究与应用,2016,37(02):16.
　Zhou Wei,Jiang Ping-fan,Li Xin.Analysis of strong rainfall caused by Typhoon “Mujigae” in 2015[J].Journal of Meteorological Research and Application,2016,37(04):16.
[9]刘英轶,陈婉华,郭晓薇.做好专业天气网站台风气象服务的几点思考[J].气象研究与应用,2017,38(04):107.
　Liu Yingyi,Chen Wanhua,Guo Xiaowei.Thoughts on improving typhoon weather service on professional weather website[J].Journal of Meteorological Research and Application,2017,38(04):107.
[10]黄荣,黄晴,屈梅芳,等.1713号台风“天鸽”外围龙卷特征及成因分析[J].气象研究与应用,2018,39(01):28.
　Huang Rong,Huang Qing,Qu Meifang,et al.Analysis of the characteristics and causes of the peripheral tornado of NO.1713 Typhoon Hato[J].Journal of Meteorological Research and Application,2018,39(04):28.

备注/Memo

备注/Memo:
收稿日期:2021-11-10。
基金项目:国家自然科学基金项目（41765002）、广西自然科学基金重点项目（2017GXNSFDA198030）、广西气象局重点基金项目：桂气科2021Z02、广西台风与海洋预报服务创新团队项目
作者简介:吴玉霜（1994—），女，助理工程师，从事专业气象服务。E-mail：1248893015@qq.com
通讯作者:黄小燕（1978—），女，博士，正研级高级工程师，从事智能计算预报技术在天气预报中的应用。E-mail：Gx＿huangxy@163.com
更新日期/Last Update: 1900-01-01