引用本文:汪宇,彭晓武,沈劲,嵇萍,邓滢,谢敏.基于气象因子的PM2.5回归预测模型研究[J].环境监控与预警,2018,10(4):8-11
ANG Yu, PENG Xiao wu, SHEN Jin, JI Ping, DENG Ying, XIE Min.Research on the Regression Model of PM2.5 Concentration Based on Meteorological Parameters[J].Environmental Monitoring and Forewarning,2018,10(4):8-11
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 1948次   下载 1019 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于气象因子的PM2.5回归预测模型研究
汪宇,彭晓武,沈劲,嵇萍,邓滢,谢敏
作者单位
汪宇1,彭晓武2,沈劲1,嵇萍1,邓滢1,谢敏1 1. 广东省环境监测中心广东 广州 510308
2. 环境保护部华南环境科学研究所广东 广州 510655 
摘要:
用Pearson相关系数分析了2013—2016年3大典型城市北京、南京和广州的ρ(PM<sub>2.5</sub>)与各气象因子的关系。结果表明,3个城市ρ(PM<sub>2.5</sub>)与各风速因子最大的相关系数依次为-0.44,-0.29和-0.37,与各气温因子最大的相关系数依次为-0.44,-0.33和-0.37,气压与南京和广州的ρ(PM<sub>2.5</sub>)正相关,气压因子最大的相关系数分别为0.25和0.34,湿度与北京ρ(PM<sub>2.5</sub>)正相关,与广州ρ(PM<sub>2.5</sub>)负相关,湿度因子最大的相关系数分别为0.49和-0.36,日照时数与北京ρ(PM<sub>2.5</sub>)相关系数为-0.46,降水量与南京和广州ρ(PM<sub>2.5</sub>)相关系数分别为-0.20和-0.24;采用逐步线性回归方法建立城市次日ρ(PM<sub>2.5</sub>)与气象因子的预测模型,复合相关系数分别为0.722 8,0.770 6和0.809 9。模型预测3个城市2016年PM<sub>2.5</sub>年均值分别偏高4,5和3μg/m<sup>3</sup>,日均值平均相对误差为±45.6%,±32.9%和±26.0%,模型对高ρ(PM<sub>2.5</sub>)普遍低估。
关键词:  细颗粒物  气象因子  相关性  线性回归  北京  南京  广州
DOI:
分类号:X513
文献标识码:B
基金项目:国家科技支撑计划基金资助项目(2014BAC21B04);国家自然科学基金面上资助项目(21477045)
Research on the Regression Model of PM2.5 Concentration Based on Meteorological Parameters
ANG Yu, PENG Xiao wu, SHEN Jin, JI Ping, DENG Ying, XIE Min
Abstract:
Using Pearson correlation coefficient, relationship between PM<sub>2.5</sub> concentration [ρ(PM<sub>2.5</sub>]and meteorological factors in three typical cities Beijing, Nanjing and Guangzhou were analyzed during 2013 to 2016. The results showed that the maximun correlation coefficient between ρ(PM<sub>2.5</sub>) and wind speed factor in the three cities was -0.44, -0.29 and -0.37 in turn, and the maximun correlation coefficient was -0.44, -0.33 and -0.37 for the temperature factor. Atmospheric pressure was positively correlated with ρ(PM<sub>2.5</sub>) in Nanjing and Guangzhou, and the maximum correlation coefficient of the pressure factor was 0.25 and 0.34, respectively. Humidity was positively correlated with ρ(PM<sub>2.5</sub>) in Beijing but negatively correlated with ρ(PM<sub>2.5</sub>)in Guangzhou, with the maximum correlation coefficient of humidity factor 0.49 and -0.36, respectively. The correlation coefficient between sunshine hours and ρ(PM<sub>2.5</sub>) of Beijing was -0.46. The correlation coefficient between precipitation and ρ(PM<sub>2.5</sub>) in Nanjing and Guangzhou was -0.20 and -0.24, respectively. The prediction model of next day ρ(PM<sub>2.5</sub>) and meteorological factors was established by the stepwise linear regression method, with the composite correlation coefficients being 0.722 8, 0.770 6 and 0.809 9 respectively. Annual average ρ(PM<sub>2.5</sub>) values in 2016 are overestimated by 4, 5 and 3 μg/m<sup>3</sup>, while the average relative errors for daily mean are ± 45.6%, ± 32.9% and ± 26.0%, respectively The model generally underestimated the high value of ρ(PM<sub>2.5</sub>).
Key words:  PM2.5  Meteorological factor  Correlation  Linear regression  Beijing  Nanjing  Guangzhou