| 基于机器学习的长江重庆航道雾情特征提取 |
| |
| 引用本文： | 陈施吉,李家启,王远谋,李科,孙恩虹.基于机器学习的长江重庆航道雾情特征提取[J].热带气象学报,2022,38(6):845-853. |
| |
| 作者姓名： | 陈施吉 李家启 王远谋 李科 孙恩虹 |
| |
| 作者单位： | 1.重庆市气象服务中心，重庆 401147 |
| |
| 基金项目： | 重庆市气象部门业务技术攻关项目YWJSGG-202135重庆市知天·智慧气象服务系统建设项目2020-500112-05-01-129096 |
| |
| 摘 要： | 基于2018—2020年长江沿线26个自动气象监测站的逐5 min能见度监测数据、重庆海事管辖水域资料和长江航道基础地理信息，利用K-Means、DTW、PCA等机器学习算法，分析了长江重庆航道雾情的时空分布、时序形态等特征。结果表明:长江重庆航道雾情过程高发区域是涪陵-忠县水域，长寿及上游水域次之; 江面雾情过程较高频率出现在夏季的6月、7月，冬季较之偏少，大部分的雾情过程时长均在1 h内，多在夜间生成及结束; 不同时间长度的雾情过程具有不同的时序形态特征，当时长不足27 h时，主要表征能见度下降过程的信号，超过27 h的过程则主要表征能见度回升阶段信号，“象鼻形”先期振荡信号随着雾情过程时长的加大而进一步增强。
|
| 关 键 词： | 重庆 长江航道 雾 机器学习 |
| 收稿时间： | 2021-05-16 |
CHARACTERISTICS OF THE FOG IN THE CHONGQING SECTION OF YANGTZE RIVER WATERWAY: A STUDY BASED ON MACHINE LEARNING |
| |
| CHEN Shiji,LI Jiaqi,WANG Yuanmou,LI Ke,SUN Enhong.CHARACTERISTICS OF THE FOG IN THE CHONGQING SECTION OF YANGTZE RIVER WATERWAY: A STUDY BASED ON MACHINE LEARNING[J].Journal of Tropical Meteorology,2022,38(6):845-853. |
| |
| Authors: | CHEN Shiji LI Jiaqi WANG Yuanmou LI Ke SUN Enhong |
| |
| Institution: | 1.Chongqing Meteorological Service Center, Chongqing 401147, China2.Changshou Meteorological Bureau, Chongqing 401220, China3.Jiangjin Meteorological Bureau, Chongqing 402260, China |
| |
| Abstract: | Based on every 5 minutes visibility monitoring data from 2018 to 2020 from 26 automatic weather monitoring stations along the Chongqing section of Yangtze River (CSYR), data of the jurisdicted water area of Chongqing Maritime Safety Service and the basic geographic information of Yangtze River, the spatiotemporal distribution and temporal morphological characteristics of the fog in the CSYR are analyzed by using machine learning algorithms such as K-means, DTW and PCA. The results show that the high incidence area of fog processes in the CSYR is the waters from Fuling to Zhongxian, followed by Changshou and its upstream waters. Fog process is more frequent in June and July, whereas less in winter. Fog usually rises and disperses at night, and the majority of fog processes has a duration within one hour. Processes with different lifespan present distinct temporal morphological characteristics. In the cases that last for less than 27 hours, signals of visibility decline are found, whereas the cases which last for more than 27 hours mainly show signals of visibility rise. In addition, the early oscillation signal that looks like elephant trunks is further enhanced with the extension of the fog process duration. |
| |
| Keywords: | Chongqing Yangtze River waterway fog machine learning |
|
| 点击此处可从《热带气象学报》浏览原始摘要信息 |
| 点击此处可从《热带气象学报》下载免费的PDF全文 |
|
