成都O_(3)逐日污染潜势关键时段优选的GAM模型  

Optimization of the key period of daily ozone pollution potential in Chengdu based on Generalized Additive Model

作　　者：任至涵 倪长健[1,2] 花瑞阳 杜云松 REN Zhi-han;NI Chang-jian;HUA Rui-yang;DU Yun-song(School of Atmospheric Sciences,Chengdu University of Information Technology,Chengdu 610225,China;Plateau Atmosphere and Environment Key Laboratory of Sichuan Province,Chengdu 610225,China;Sichuan Ecological Environment Monitoring Station,Chengdu 610091,China)

机构地区：[1]成都信息工程大学大气科学学院,四川成都610225 [2]高原大气与环境四川省重点实验室,四川成都610225 [3]四川省生态环境监测总站,四川成都610091

出　　处：《中国环境科学》2021年第11期5079-5085,共7页China Environmental Science

基　　金：国家重点研发计划项目(2018YFC0214004,2018YFC1506006);四川省科技厅应用基础研发项目(2021YJ0314)。

摘　　要：利用成都市2016~2018年O_(3)逐时监测数据以及该时段同时次的地面气象观测资料,通过对O_(3)日变化特征的分析,确定了表征研究区O_(3)逐日污染潜势的四个关键时段,即全天时段(00:00~24:00)、日间时段(05:00~20:00)、O_(3)超标时段(11:00~19:00)以及O_(3)峰值时段(15:00~16:00).基于广义可加模型(Generalized Additive Model,GAM)分别构建了O_(3)日最大8h滑动平均浓度(O_(3-8h))与上述四个时段气象要素之间的函数关系,分析了时间尺度变化对O_(3)逐日污染潜势的影响.结果表明:GAM模型可以很好地表征O_(3-8h)与不同时段多气象要素之间的非线性关系.O_(3)超标时段气象要素对O_(3)逐日污染潜势具有最佳的指示意义,对应GAM模型的调整判定系数R^(2)和方差解释率IRV分别为0.81和81.4%,模型模拟值与观测值的压轴回归决定系数R^(2)为0.805.太阳辐射、相对湿度和气温是决定O_(3)逐日污染潜势最重要的气象要素,但三者在GAM模型中的重要性排序会因时间尺度的变化而有所差异.Daily characteristics of O_(3) were analyzed using hourly observational data of O_(3) concentration and corresponding meteorological data from 2016 to 2018 in Chengdu.Four key periods featuring daily ozone pollution potential were selected which were all day period(0:00~24:00),daytime period(05:00~20:00),the period of O_(3) exceeding the standard(11:00~19:00)and the period of O_(3) reaching its peak(15:00~16:00).The relationship between the maximum 8-hour moving average of O_(3)(O_(3-8 h))and meteorological factors during the four periods were established based on Generalized Additive Model(GAM).The effects of the variation of time scale on daily ozone pollution potential were also investigated.The results showed that GAM model could simulate and capture the non-linear relationship between O_(3-8 h) and meteorological factors.Meteorological factors during the period of O_(3) exceeding the standard(from 11:00 to 19:00)could best represent daily O_(3) pollution potential with adjusted determination coefficient R^(2) and interpretation rate of variance of 0.81 and 81.4%,respectively.The determination coefficient R^(2) of reduced major axis regression(RMA)between simulated O_(3)-8 and observed ones was 0.805.Solar radiation,relative humidity and air temperature,were three main factors affecting daily O_(3) pollution potential.However,the ranking of the importance varied with time scale in the certain one of the four GAM models.

关 键 词：臭氧 逐日污染潜势 关键时段 GAM模型 成都 

分 类 号：X515[环境科学与工程—环境工程]