基于随机森林模型的PM_(2.5)成分NO_(3)^(-)浓度估算  

Estimation of the concentration of PM_(2.5)-bound composition NO_(3)^(-) based on random forest model

作　　者：周珍 张亚一 王情[2] 高祥伟 马润美 班婕[2] 陆开来 ZHOU Zhen;ZHANG Ya-yi;WANG Qing;GAO Xiang-wei;MA Run-mei;BAN Jie;LU Kai-lai(School of Marine Technology and Geomatics,Jiangsu Ocean University,Lianyungang 222005,China;National Institute of Environmental Health,Chinese Center for Disease Control and Prevention)

机构地区：[1]江苏海洋大学海洋技术与测绘学院,连云港222005 [2]中国疾病预防控制中心环境与健康相关产品安全所

出　　处：《环境卫生学杂志》2022年第3期177-183,共7页JOURNAL OF ENVIRONMENTAL HYGIENE

基　　金：国家自然科学基金面上项目(42071433);国家自然科学基金青年基金项目(41701234)。

摘　　要：目的以硝酸根离子(NO_(3)^(-))为例,建立基于随机森林算法的PM_(2.5)成分浓度估算模型,并获得对NO_(3)^(-)浓度影响较大的因子,以及NO_(3)^(-)浓度的连续时间序列特征。方法研究以2013—2017年气象、土地利用、排放清单和PM_(2.5)、NO_(2)、PM_(10)、SO_(2)、CO空气质量监测数据为自变量,以NO_(3)^(-)浓度数据为因变量,利用值提取至点、反距离权重插值和设置1 km缓冲区等方法将各类数据集标准化。构建随机森林模型,并采用十折交叉法对模型拟合效果进行验证。结果模型验证结果表明,模拟值和真实值的拟合程度较高,日均、月均和年均浓度R^(2)分别为0.61,0.77和0.83。由NO_(3)^(-)浓度的模型特征参数重要性排序可得,PM_(2.5)质量浓度的重要性得分最高(0.387),反照率滞后2日(lag2)、反照率滞后1日(lag1)、10 m经向风速,边界层高度等气象因素与NO_(3)^(-)浓度变化关系较密切。此外,交通、民用、工业和电力部门排放的一次PM_(2.5)源均排在重要性前20名。结论多参数的随机森林模型在PM_(2.5)成分模拟中有一定的优越性;PM_(2.5)质量浓度、NO_(2)、10 m经向风速、生活源和交通源的一次PM_(2.5)源等因子对于NO_(3)^(-)浓度模拟影响较大;NO_(3)^(-)浓度存在一定的季节分布特征。Objective To establish a PM_(2.5) component concentration estimation model based on random forest algorithm with NO_(3)^(-) as an example,and to investigate the large influencing factors for NO_(3)^(-) concentration and the continuous time series characteristics of NO_(3)^(-) concentration.Methods The study used the meteorological,land use,emission inventory,and air quality monitoring data of PM_(2.5),NO_(2),PM_(10),SO_(2) and CO between 2013 and 2017 as the independent variables and NO_(3)^(-) concentration data as the dependent variable,and various method such as value extraction to points,inverse distance weight interpolation,and setting of 1 km buffer area were used to standardize various data sets.A random forest model was established,and the fitting effect of the model was validated by the ten-fold crossover method.Results The result of model verification showed that there was a high degree of fitting between the simulated value and the monitoring value,and daily,monthly,and annual mean concentrations had R^(2) of 0.61,0.77,and 0.83,respectively.According to the importance ranking of the feature parameters of NO_(3)^(-) concentration in the model,the mass concentration of PM_(2.5) had the highest importance score of 0.387,and meteorological factors such as albedo lagging for 2 days,albedo lagging for 1 day,10 m longitudinal wind speed and height of the boundary layer were closely associated with the change in NO_(3)^(-) concentration.In addition,the primary PM_(2.5) sources emitted by transportation,residential,industry,and power sectors all ranked among the 20 most important sources.Conclusion The multi-parameter random forest model has certain advantages in PM_(2.5) composition simulation.Factors such as PM_(2.5) mass concentration,NO_(2),10 m longitudinal wind speed,and primary PM_(2.5) sources emitted by residential and traffic sectors have a great influence on the simulation of NO_(3)^(-) concentration.NO_(3)^(-) concentration has the characteristics of seasonal distribution.

关 键 词：细颗粒物(PM_(2.5)) 成分 硝酸根离子(NO_(3)^(-)) 随机森林 浓度模拟 

分 类 号：X511[环境科学与工程—环境工程] X820