基于小波分析的太原市空气质量变化特征及预测  

Variation and Prediction of Air Quality in Taiyuan Based on Wavelet Analysis

作　　者：姚清晨 张红[2] YAO Qingchen;ZHANG Hong(Taiyuan Environmental Monitoring Central,Taiyuan 030002,China;School of Environmental &Resource Sciences,Shanxi University,Taiyuan 030006,China)

机构地区：[1]太原市环境监测中心站,山西太原030002 [2]山西大学环境与资源学院,山西太原030006

出　　处：《山西大学学报（自然科学版）》2019年第1期265-274,共10页Journal of Shanxi University(Natural Science Edition)

基　　金：山西省自然科学基金(201601D202086)

摘　　要：以太原市2013年1月～2016年12月份PM2.5、NO_2、SO_2等污染物逐日浓度数据为研究对象,结合太原市地面气象数据,采用相关分析、小波分析等方法对太原市空气质量AQI(air quality index)变化特征进行了研究,同时采用小波去噪和最优子集回归方法分别建立AQI的春、夏、秋、冬季预报方程。研究结果表明:1)太原市AQI均值呈现逐年降低趋势,最大值出现在冬季,具有冬强夏弱的特点,太原市主要空气污染物为PM2.5,PM10和SO_2。2)AQI与各污染物浓度因子之间存在较强的相关性,其中AQI与PM2.5和PM10的相关性最大,Spearman相关系数极显著(P<0.01),并且污染物之间、污染物与气象因子之间也存在相关性。3)太原市AQI具有较明显的年际周期性振荡、30～60d的季节性周期振荡、10～20d的双周性振荡及5～7d的准双周振荡。4)将AQI前一天的历史数据作为因子引入预测模型,相比于仅以气象因素为输入的模型具有更强的拟合精度。对数据进行小波去噪后所建的最优子集回归方程比使用原始数据更优。文章所建立的"去噪气象数据+去噪历史AQI数据"模型可以较精确地实现对太原市AQI指数的短期预测。Based on the air pollutants monitoring data and meteorological data from January 2013 to December 2015 in Taiyuan,the variation characters of air quality(AQI)of Taiyuan were analyzed by using correlation analysis and wavelet analysis.The optimal subset regression method was used to establish the prediction equation of AQI in spring,summer,autumn and winter respectively.The results show that:1)The AQI of Taiyuan decreased over the years,and appeared the similar seasonal tendency with high in winter and low in summer.The PM2.5,PM10 and SO2 were the main pollutants in Taiyuan.2)There was a strong correlation between AQI and various pollutant concentration factors,the Spearman correlation coefficients between AQI with PM2.5and PM10 were significant(P<0.01)respectively.The AQI was also significantl correlated with meteorological factors(P<0.05).3)AQI had obvious interannual periodic oscillations,seasonal oscillations of 30-60days,biweekly oscillations of 10-20days and quasi-biweek oscillation of 5-7days.4)The prediction equations which include the history AQI and meteorological factors together had better fitting effects than the equations which only include meteorological factors.The prediction equations based on the wavelet de-noising data were better than that of the original data.The"de-noising meteorology+de-noising history data"model can accurately predict the short term AQI of Taiyuan.

关 键 词：空气质量指数 时间序列分析 小波分解 小波去噪 最优子集回归 

分 类 号：X51[环境科学与工程—环境工程]