主办单位:中国气象局沈阳大气环境研究所
国际刊号:ISSN 1673-503X
国内刊号:CN 21-1531/P

气象与环境学报 ›› 2019, Vol. 35 ›› Issue (2): 15-22.doi: 10.3969/j.issn.1673-503X.2019.02.003

• 论文 • 上一篇    下一篇

基于多元线性回归的滇西南短时强降水预报模型研究

王秀英1, 廖留峰2,3, 王俊杰4   

  1. 1. 普洱市气象局, 云南 普洱 665000;
    2. 贵州省山地环境气候研究所, 贵州 贵阳 550002;
    3. 贵州省山地气候与资源重点实验室, 贵州 贵阳 550002;
    4. 普洱学院, 云南 普洱 665000
  • 收稿日期:2018-04-02 修回日期:2018-07-24 出版日期:2019-04-30 发布日期:2019-04-30
  • 作者简介:王秀英,女,1986年生,工程师,主要从事预报技术方面研究,E-mail:410518631@qq.com
  • 基金资助:

    贵州省科技厅项目(黔科合支撑2017(2593))和云南省普洱学院创新团队(CXTD003)项目共同资助。

A forecast model for flash heavy rainfall in southwestern Yunnan province based on a multiple linear regression method

WANG Xiu-ying1, LIAO Liu-feng2,3, WANG Jun-jie4   

  1. 1. Puer Meteorological Office, Puer 665000, China;
    2. Guizhou Institute of Mountainous Environment and Climate, Guiyang 550002, China;
    3. Guizhou Key Laboratory of Mountainous Climate and Resources, Guiyang 550002, China;
    4. Puer University, Puer 665000, China
  • Received:2018-04-02 Revised:2018-07-24 Online:2019-04-30 Published:2019-04-30

摘要:

利用云南省普洱市2015—2017年多普勒天气雷达资料、探空资料和气象观测站5 min雨量观测资料,分析了普洱地区研究期间41次短时强降水的环境场和雷达回波演变特征。结果表明:中尺度辐合线、中气旋、逆风区是强降水触发和维持的重要成因。短时强降水发生前,整层大气水汽充沛,静力不稳定层结,大气可降水量(PW)≥35 mm、SI≤-0.23、K>35,可作为环境场对流潜势的判定因子;短时强降水发生时,雷达回波最强反射率因子≥40 dBz,35 dBz回波顶高>5 km,径向速度的辐合切变量>5 m·s-1。通过多元线性回归分析,选取4个相关性显著的影响因子,建立普洱市短时强降水预报模型。所选预报因子包括:35 dBz回波顶高、30 dBz垂直剖面中心高度、30 dBz以上雷达回波面积和SI。预报模型的回报检验表明,普洱短时强降水平均雨强相对均方根误差为17.0%,局地降水持续时间相对均方根误差为33.9%,局地过程降水相对均方根误差为25.6%,回报效果较好。4次短时强降水预报检验中,平均雨强的预报误差每5 min小于1.2 mm,局地强降水持续时间的预报误差小于10 min,局地过程降水的预报误差小于4 mm,模型均预报出局地连续性降水超过50 mm。预报模型有较好的预报能力,可应用于普洱短时强降水的临近预报预警。

关键词: 短时强降水, 临近预报, 预报因子, 多元线性回归, 预报检验

Abstract:

Using the data from a Doppler weather radar,the sounding data,and 5-min precipitation from meteorological stations in Puer,the evolution characteristics of environmental fields and radar echoes during 41 flash heavy rainfall events from 2015 to 2017 were analyzed.The results showed that mesoscale convergence lines,mesocyclones and adverse wind zones are important causes for the triggering and maintenance of flash heavy rainfall events.Before flash heavy rainfall occurs,the entire atmosphere has abundant water vapor and statically unstable stratification,with precipitable water (PW) ≥ 35 mm,a Showalter index (SI) ≤-0.23,and a K index >35,which can be used as determining factors of the convection potential in the environmental fields.When flashing heavy rainfall occurs,the strongest reflectance factor of radar echoes reaches 40 dBz at least,the echo tops of 35 dBz (RHI35) is larger than 5 km,and the radial velocity of convergence shear variable is larger than 5 m·s-1.A forecast model for flash heavy rainfall in Puer is established using four influencing factors with significant correlation selected through the multiple linear regression analysis,including RHI35,vertical section center height of 30 dBz,radar echo area more than 30 dBz,and the SI.The validation of this model shows that the relative root mean square (RMS) error of average rainfall intensity is 17.0%,and those of precipitation duration and the amount in local processes are 33.9% and 25.6%,respectively.The forecast error of average rainfall intensity in four flash heavy rainfall events is less than 1.2 mm/5 min,and that for precipitation duration and amount is less than 10 min and 4 mm,respectively.The consistent precipitation exceeds 50 mm for the four events,which is also captured in the model.Overall,the forecast model has good performance and can be used for the short-time forecast of flash heavy rainfall events in Puer area.

Key words: Flash-heavy rain, Nowcasting, Forecast factor, Multiple linear regression, Forecast verification

中图分类号: