J4 ›› 2012, Vol. 30 ›› Issue (3): 459-464.

• 业务技术应用 • 上一篇    下一篇

年降水量数据的正态变换方法对比分析

陈学君1 ,苏仲岳2 ,李仲龙1 ,韩涛3   

  1. 1. 甘肃省气象信息与技术装备保障中心,甘肃兰州730020;
    2. 兰州大学数学与统计学院,甘肃兰州730000; 3. 西北区域气候中心,甘肃兰州730020
  • 收稿日期:2012-03-02 出版日期:2012-09-30 发布日期:2012-05-22
  • 作者简介:陈学君( 1973 - ) ,男,甘肃天水人,博士,高级工程师,主要从事数据分析与处理、气象数据质量控制工作. E - mail: xuejunchen1971@163. com
  • 基金资助:

     国家公益性行业( 气象) 科研专项项目“中国近60 年地面关键气候要素均一性检验与订正技术及站址变动影响研究”
    ( GYHY201206013) 、科技部农业科技成果转化资金项目“西北区域农业干旱监测预警技术推广应用”( 2011GB24160005) 、甘肃
    省气象局科研项目“基于多源卫星资料的甘肃省雪盖及雪深监测方法研究”( 2011 - 02) 共同资助
     

Comparative Analysis of the Normal Transformation Methods about Annual Precipitation

CHEN Xuejun1,SU Zhongyue2,LI Zhonglong1,HAN Tao3   

  1.  1. Meteorological Information & Technique Support & Equipment Centre
    of Gansu Province,Lanzhou 730020,China; 2. College of Mathematics and Statistics,
    Lanzhou University,Lanzhou 730000,China; 3. Regional Climate Center,Lanzhou 730020,China
  • Received:2012-03-02 Online:2012-09-30 Published:2012-05-22

摘要:

对于长时间尺度的降水序列变异性强、偏度大、不符合正态分布年降水数据,直接进行非均一性检验、气候统计分析等处理会产生较大误差,需首先选择合适的正态变换方法进行稳健处理。以甘肃省定西、敦煌、武都、镇原4 个气象站点年降水资料为例,对于不服从正态分布的站点分别采用Box- Cox 变换和Johnson 变换进行正态变换。对比结果显示Box - Cox 变换与Johnson 变换使数据接近正态分布均是有效的。其中, Johnson 变换对于异常数据的正态变换效果比Box - Cox 变换更优。

关键词: 甘肃省, 降水量, 正态变换, 统计检验

Abstract:

For the annual precipitation data,non - homogeneity test and climate statistical analysis directly will lead to considerable errors,
so at first it needs to choose appropriate normal transform method. In this paper,based on precipitation data from four typical stations
( Dingxi,Dounhuang,Zhenyuan and Wudu station ) in Gansu province,the normal transform methods about Box - Cox transformation
and Johnson transform to non - normal data were adopted. The result shows that both the Box - Cox transformation and Johnson
transformation made precipitation data close to normal distribution,and the Johnson transformation was better than the Box - Cox transformation
about normal transformation effect to abnormal data.

Key words: Gansu province, precipitation, normal transformation

中图分类号: