基于轮廓似然函数的PM_(10)极值浓度分析  

Application of Analyzing the Extreme Value PM_(10) Base on Profile Likelihood Function

作　　者：李柚[1] 王斌[1] 杨育杰[1] 

机构地区：[1]四川大学建筑与环境学院,四川成都610065

出　　处：《中国环境监测》2010年第6期31-35,共5页Environmental Monitoring in China

摘　　要：以R软件为分析工具,选择GEV(generalized extreme value distribution)模型拟合四川省泸州市2003～2007年期间PM10每月最高日平均浓度数据,采用极大似然法估计模型的3个参数即位置参数、尺度参数、形状参数,利用所得的参数估计值计算得出某一标准值(如GB3095—1996)的重现期;进一步利用参数估计值计算轮廓似然函数,估计某一段固定时间间隔的PM10浓度的重现值以及其置信区间。结果表明,GEV模型能很好地拟合泸州市PM10数据,利用轮廓似然函数估计的不同时间间隔的重现值准确度高,统计结果可以为环境主管部门发布污染状况预警信息提供参考。Use R software as a analytical tool,the GEV（generalized extreme value distribution） model has been used to fit the parent distribution of the monthly maximum PM10 data of Lu Zhou Si Chuan from 2003 to 2007.It has been found that generalized extreme value distribution represent these air quality data very well.The three parameters of the generalized extreme value distribution（GEV）,location parameter,scale parameter,shape parameter,have been estimated by maximum likelihood method.Furthermore,used the parameter estimation results to calculate the profile likelihood function values,the results were taken to estimate the return period and exceedances of a critical PM10 concentration,such as China’s ambient air quality standard（GB3095—1996）,and also a fixed interval’s return value and it’s confidence interval could be obtain by this process.We can see,the GEV model is a good tool to analyze the extreme PM10 data and the statistical results can be used as pollution early-warning information.

关 键 词：R软件 GEV分布 重现期 轮廓似然函数 

分 类 号：X830.3[环境科学与工程—环境工程]