胡海川,张恒德,朱彬,谢超.神经网络方法在环渤海能见度预报中的应用分析.气象科学,2018,(6):798-805 HU Haichuan,ZHANG Hengde,ZHU Bin,XIE Chao.Application analysis of neural network method in visibility forecast of coastal cities around Bohai Sea.Journal of the Meteorological Sciences,2018,(6):798-805
神经网络方法在环渤海能见度预报中的应用分析
Application analysis of neural network method in visibility forecast of coastal cities around Bohai Sea
投稿时间:2017-11-22  修订日期:2018-05-08
DOI:10.3969/2017jms.0041
中文关键词:  集合预报  神经网络  能见度  预报技术
英文关键词:Ensemble forecast  Neural network  Visibility  Forecasting technique
基金项目:国家重点研发计划课题(2016YFC0203301);国家基金委重点研究项目(91644223)
作者单位E-mail
胡海川 南京信息工程大学, 南京 210044
国家气象中心, 北京 100081 
 
张恒德 国家气象中心, 北京 100081 zhanghengde1977@163.com 
朱彬 南京信息工程大学, 南京 210044  
谢超 国家气象中心, 北京 100081  
摘要点击次数: 102
全文下载次数: 201
中文摘要:
      本文基于2001-2015年中国气象局地面常规气象观测数据及NECP再分析资料,采用BP神经网络方法构建环渤海沿海城市能见度预报模型,利用2016年ECMWF集合预报数据基于预报模型进行能见度预报实验,并与ECMWF集合预报产品中现有能见度预报结果进行对比分析。分析表明:该方法对于环渤海沿海城市能见度预报的预报效果明显高于ECMWF集合预报中的能见度预报,12~72 h预报时效中,最小值对应1 km以下能见度的TS评分为0.36~0.43;10 km以下能见度预报误差显著降低,与离散度的对应关系较好。因此,该方法对低能见度天气过程的能见度预报具有指示意义。
英文摘要:
      Based on the conventional ground-based observation data of China Meteorological Administration during the period of 2001 to 2015 and NECP reanalysis data, and by using the BP neural network method, this paper constructs the forecast model of visibility in the coastal cities around Bohai Sea, to conduct a visibility forecast experiment based on the forecast model and by using the 2016 ECMWF ensemble forecast data, and compare the result to the current visibility forecast result from products of ECMWF ensemble forecast. The analysis shows distinct improvement of this method in the accuracy of visibility forecast in the coastal cities around Bohai sea compared to visibility forecast of ECMWF ensemble forecast. In 12-72 hours of forecast periods, the minimum value corresponds to the TS rated score when visibility is below 1km, which is 0.36-0.43; it also shows a significant reduction of deviation in visibility forecast when visibility is below 10 km, with good dispersion relation as well. Also, it can provide an important reference for visibility forecast of low visibility weather.
查看全文  查看/发表评论  下载PDF阅读器
关闭