Journal of Arid Meteorology

Previous Articles     Next Articles


Probability Forecast Method of Thunderstorm in East Region of Northwest China Based on Stepwise Regression Analysis

KONG Debing 1,2, SHANG Kezheng 1, WANG Shigong 1 , ZHAO Wenjing 1, YE Wei 2
  

  1. 1. Key Laboratory of Semi-Arid Climate Change of Ministry of Education,College of Atmospheric Sciences, Lanzhou University, Lanzhou 730000,China;
    2. The Meteorological Observatory of PLA 95605 Unit, Chongqing 402361,China
  • Online:2016-02-29 Published:2016-02-29

基于逐步回归分析的西北地区东部雷暴概率预报方法研究

孔德兵1,2尚可政1王式功1赵文婧1叶伟2   

  1. 1.兰州大学大气科学学院/半干旱气候变化教育部重点实验室,甘肃兰州730000;
    2.中国人民解放军95605部队气象台,重庆402361
  • 通讯作者: 尚可政(1960-),男,甘肃景泰人,博士,硕士生导师,主要从事干旱气候和现代天气预报技术和方法研究.E-mail:shangkz@lzu.edu.cn
  • 基金资助:

    国家公益性(气象)行业专项项目(GYHY201206004、GYHY201306047)和兰州大学中央高校基本科研业务费专项(lzujbky-2013-m03)共同资助

     

Abstract:

Based on the weather circumstances data of 60 meteorological sites in east region of Northwest China during 2008-2013 and convective parameters calculated from NCEP/NCAR data, the thunderstorm weather circumfluence types were classified, the samples that did not conform to the thunderstorm weather types were eliminated firstly, then the event probability regression method was used to set up the April-October 24 h thunderstorm potential forecast formulas, and on this basis, the thunderstorms occurring in 2013 were forecasted. The results are as follows: (1) Thunderstorms mostly occurred in the northeast slope of the Qingzang Plateau and rarely occurred in the Gobi Desert in the north and the Central Shaanxi Plain in the east, and the annual average thunderstorm days were five to ten days in 45% sites of all sites in east region of Northwest China. (2) In the thunderstorm weather circumfluence classification processes, a lot of samples in which thunderstorms did not happen were successfully reduced in order to improve forecast precision. (3) The meteorological convective parameters calculated from NCEP/NCAR data were highly related to the occurrence of thunderstorms and had a larger variance contribution to the thunderstorm potential formulas. (4) The average TS score for 60 sites during 2008-2012 was 24.2%, and the average TS score for 60 sites in 2013 was 23.3%. The thunderstorm potential formulas had better forecast effect on sites where more thunderstorms happened. The forecast result is reasonable and can provide reference for the study of thunderstorm in east region of Northwest China.

Key words:  thunderstorm, stepwise regression, probability forecast

摘要:

利用2008~2013年西北地区东部60个站点天气实况资料以及NCEP/NCAR 再分析资料计算出的对流参数,统计各站年均雷暴日数,在前期天气分型的基础上,通过事件概率回归方法,建立了60个站点4~10月24 h雷暴概率预报方程,并对2013年4~10月各站雷暴进行试预报。结果表明:(1)西北地区东部雷暴发生最多的站点位于青藏高原东北边坡地带,最少的站点位于北部戈壁荒漠地带与东部关中平原一带,45%的站点年均雷暴日数在5 ~10 d之间;(2)前期天气分型过程剔除了一部分未发生雷暴的样本,使得逐步回归建模的总样本数明显减少,有效地提高了预报准确率;(3)基于NCEP/NCAR再分析资料计算的物理量与雷暴有无事件相关性较好,对24 h概率回归方程方差贡献较大;(4)2008~2012年回代预报所有站点平均TS评分为24.2%;2013年试预报所有站点平均TS评分为23.3%,雷暴发生较多的站点预报效果更好。该预报结果较理想,可为西北地区东部雷暴研究提供参考。

关键词: 雷暴, 逐步回归, 概率预报

CLC Number: