中国农业气象 ›› 2014, Vol. 35 ›› Issue (06): 690-699.doi: 10.3969/j.issn.1000-6362.2014.06.013

• 论文 • 上一篇    下一篇

非线性克里格方法在日降水量分布估算中的应用

张唯,郑海波,张剑波   

  1. 中国地质大学(武汉)信息工程学院,武汉430074
  • 收稿日期:2014-04-02 出版日期:2014-12-20 发布日期:2015-05-21
  • 作者简介:张唯(1980-),女,湖北云梦人,讲师,博士,主要研究领域为空间统计分析、数字地形分析。
  • 基金资助:
    国家自然科学基金青年科学基金(41001225);中央高校基本科研业务费专项资金(CUGL120272)

Application of Nonlinear Kriging Method on Estimation of Daily Precipitation Distribution

ZHANG Wei,ZHENG Hai bo,ZHANG Jian bo   

  1. Faculty of Information Engineering,China University of Geosciences,Wuhan430074,China
  • Received:2014-04-02 Online:2014-12-20 Published:2015-05-21

摘要: 为研究非正态分布条件下的日降水量空间分布,选择湖北省75个气象观测站2010年7月的降水过程资料,通过正态变换和变差函数分析,引入多元高斯克里格方法进行逐日降水量空间分布估算,并与其它方法进行对比分析。结果表明:(1)通过合理的各向异性参数设置及理论变差函数选择,能够充分凸显降水量分布特征。(2)对比发现,反距离加权法(IDW)、普通克里格法(OK)和多元高斯克里格法(MGK)3种估算方法中,MGK估算的平均误差最小且误差曲线最平稳,是该地区非正态分布日降水量较适合的空间分布估算方法。(3)使用MGK方法,能有效改善IDW带来的“牛眼效应”以及OK法造成的过度平滑,得到具有较高精度的日降水量数据集。

关键词: 非正态分布, 多元高斯克里格, 变差函数分析, 日降水量

Abstract: In order to study spatial distribution of the nonnormal precipitation,the spatial estimation of daily precipitation was performed and the precision was comparatively validated by using of the MultiGaussian Kriging Method(MGK),through normal transformation and variogram analysis for the precipitation process data form 75 meteorological stations in Hubei province in July 2010.The results showed that the characteristic of precipitation distribution was highlighted through the reasonable parameter selection,and the estimation accuracy also improved significantly.MGK method was the most suitable method for spatial estimation of nonnormal precipitation,with minimal average mean error and most stable error curve,among three interpolation methods of Inverse Distance Weighted,MultiGaussian Kriging and Ordinary Kriging.The "bullseye effect"caused by IDW and the overly smooth caused by OK had been effectively improved under Multi Gaussian Kriging method,and the daily precipitation spatial distribution dataset with higher precision could be obtained.

Key words: Non normal distribution, Multi Gaussian Kriging, Variogram analysis, Daily precipitation