摘要
传统的地统计学方法需要区域降水量满足多元正态假设才能给出无观测站位置降水量的概率密度函数,经典的统计学方法则用唯一的概率密度函数对区域内任一无观测站位置的降水量进行统计描述。尝试用基于Copula的地统计学方法对无观测站位置月降水量的不确定性进行建模,并与上述两种方法进行对比分析。案例研究表明,基于Copula的地统计学方法可以得到任一无观测站位置的概率密度函数,且不同位置的概率密度函数与其周围观测站的分布及其观测值有关;基于Copula的地统计学方法交叉验证的区间覆盖率优于普通克里格和基于Box-cox变换的普通克里格。
The traditional geostatistical model requires regional precipitation fulfills assumption of multivariate normal distribution in order to estimate the probability density function （PDF） of any ungauged location. And the classical statistical model describes precipitation of any ungauged location using only one PDF. The study attempted to use copula-based geostatistical technology to model the uncertainty of monthly precipitation at any ungauged location and compare it with the above two methods. A case study showed that Copula-based geostatistical model could get the PDF of any ungauged location, which not only depended on the density of the observation network, but also on the magnitude of the measurements and the coverage of cross-validation confidence intervals was better than that of the ordinary Kriging and ordinary Kriging with Box-Cox transformation.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2015年第6期805-809,共5页
Geomatics and Information Science of Wuhan University
基金
公益性行业(气象)科研专项资助项目(GYHY201206019)~~
关键词
COPULA
降水量
插值
不确定性
概率密度函数
Copula
precipitation
interpolation
uncertainty
probability density function
