张兵, 沈帆, 陈楠, 全继宏, 操文祥, 郑明明, 刘虹, 田一平, 袁晶. 偏最小二乘回归法研究气象因素对武汉市空气质量参数的影响[J]. 环境与职业医学, 2014, 31(4): 241-246. DOI: 10.13213/j.cnki.jeom.2014.0057
引用本文: 张兵, 沈帆, 陈楠, 全继宏, 操文祥, 郑明明, 刘虹, 田一平, 袁晶. 偏最小二乘回归法研究气象因素对武汉市空气质量参数的影响[J]. 环境与职业医学, 2014, 31(4): 241-246. DOI: 10.13213/j.cnki.jeom.2014.0057
ZHANG Bing , SHEN Fan , CHEN Nan , QUAN Ji-hong , CAO Wen-xiang , ZHENG Ming-ming , LIU Hong , TIAN Yi-ping , YUAN Jing . Relationship between Meteorological Factors and Parameters of Air Quality in Wuhan by Partial Least Squares Regression[J]. Journal of Environmental and Occupational Medicine, 2014, 31(4): 241-246. DOI: 10.13213/j.cnki.jeom.2014.0057
Citation: ZHANG Bing , SHEN Fan , CHEN Nan , QUAN Ji-hong , CAO Wen-xiang , ZHENG Ming-ming , LIU Hong , TIAN Yi-ping , YUAN Jing . Relationship between Meteorological Factors and Parameters of Air Quality in Wuhan by Partial Least Squares Regression[J]. Journal of Environmental and Occupational Medicine, 2014, 31(4): 241-246. DOI: 10.13213/j.cnki.jeom.2014.0057

偏最小二乘回归法研究气象因素对武汉市空气质量参数的影响

Relationship between Meteorological Factors and Parameters of Air Quality in Wuhan by Partial Least Squares Regression

  • 摘要: 目的 用偏最小二乘回归法研究2009-2012年武汉市可吸入颗粒物(PM10)、二氧化硫(SO2)和二氧化氮(NO2)浓度的气象影响因素。

    方法 研究数据为2009-2012年武汉市环境自动监测点监测的PM10、SO2 和NO2 日平均浓度,以及同期武汉市地面气象观测台所观测的24 h(当日20:00到次日20:00)累积降水量、平均气压、平均风速、平均气温、平均相对湿度和平均日照时数。用偏最小二乘回归法提取变量的变异信息,交叉验证方法确定最佳主成分数,进而确定气象因素与大气污染物间的线性关系。

    结果 2009-2012年间武汉市空气质量逐年改善,呈现PM10 和NO2 复合型大气污染特征。偏最小二乘回归分析结果提示,第一主成分对PM10、SO2 和NO2 变异的解释能力分别为0.722、0.915和0.702。24 h 累积降水量的增加、平均风速的加大、平均气温的升高、平均相对湿度和平均日照时数的增加能够降低该城空气质量参数(PM10、SO2 和NO2)的浓度,而平均气压的上升会使PM10、SO2 和NO2 浓度增加。气象因素对PM10、SO2 和NO2 浓度的作用程度存在差异。平均风速和平均温度是气态污染物浓度的主要气象影响因子,降水量对NO2 浓度的稀释作用明显强于对PM10浓度的影响。

    结论 偏最小二乘回归法能够克服气象因素之间的多重共线性。气象因素(降水量、平均风速、平均气温、平均相对湿度、平均日照时数、平均大气压)对武汉市大气污染物浓度的影响存在差异性。

     

    Abstract: Objective To assess the relationship between meteorological factors and selected air quality indicators including concentrations of particulate matters with particle size below 10 microns (PM10), sulfur dioxide (SO2), and nitrogen dioxide (NO2) in Wuhan during 2009-2012 by partial least squares regression (PLSR).

    Methods Data on daily concentrations of PM10, SO2, and NO2 in Wuhan during 2009-2012 were provided by Hubei Environment Monitoring Central Station. During the same time period, the meteorological factors including 24-hour accumulated precipitation, mean pressure, mean wind speed, mean temperature, average relative humidity, and average daily hours of sunshine were synchronously obtained from China Meteorological Data Sharing Service System. Linear equations were established using the selected meteorological factors and air pollutants based on PLSR to extract variable variations, and a cross-validation approach was then applied to determine the number of principal components.

    Results The air quality in Wuhan was gradually improved from 2009 to 2012. A combined PM10 and NO2 pollution pattern was observed. The results of PLSR analysis indicated that the scores of the first principal component to PM10, SO2, and NO2 were 0.722, 0.915, and 0.702, respectively. The increases of 24-hour accumulated precipitation, mean wind speed, mean temperature, average relative humidity, and average daily hours of sunshine were along with decreasing ambient concentrations of PM10, SO2,and NO2. Contrarily, the mean pressure showed an opposite effect on the above ambient concentrations. However, the contributions of meteorological factors to the atmospheric concentrations of PM10, SO2, and NO2 varied. The mean wind speed and the mean temperature were the leading meteorological factors affecting gaseous pollutants. The dilution effect of rainwater on atmospheric NO2 concentrations was stronger than that on PM10 concentrations.

    Conclusion PLSR is able to overcome the multicollinearity of meteorological factors. The effects of meteorological factors in Wuhan, including 24-hour accumulated precipitation, mean pressure, mean wind speed, mean temperature, average relative humidity, and average daily hours of sunshine, are varied on atmospheric concentrations of PM10, SO2, and NO2.

     

/

返回文章
返回