基于多种新型受体模型的PM_(2.5)来源解析对比  

PM_(2.5) Source Apportionment Based on a Variety of New Receptor Models

作　　者：王振宇 李永斌 郭凌 宋志强 许艳玲 王丰 梁维青 史国良[1,2] 冯银厂 WANG Zhen-yu;LI Yong-bin;GUO Ling;SONG Zhi-qiang;XU Yan-ling;WANG Feng;LIANG Wei-qing;SHI Guo-liang;FENG Yin-chang(State Environmental Protection Key Laboratory of Urban Ambient Air Particulate Matter Pollution Prevention and Control,College of Environmental Science and Engineering,Nankai University,Tianjin 300350,China;China Meteorological Administration-Nankai University(CMA-NKU)Cooperative Laboratory for Atmospheric Environment-Health Research,Tianjin 300074,China;Jinzhong Municipal Bureau of Ecology and Environment,Jinzhong 030600,China;Center for Regional Air Quality Simulation and Control,Chinese Academy for Environmental Planning,Beijing 100012,China)

机构地区：[1]南开大学环境科学与工程学院国家环境保护城市空气颗粒物污染防治重点实验室,天津300350 [2]中国气象局-南开大学大气环境与健康研究联合实验室,天津300074 [3]晋中市生态环境局,晋中030600 [4]生态环境部环境规划院区域空气质量模拟与管控研究中心,北京100012

出　　处：《环境科学》2022年第2期608-618,共11页Environmental Science

基　　金：国家自然科学基金项目(41775149,42077191);中央高校基本科研业务费专项(63213072);大气重污染成因与治理攻关项目(DQGG-05-30)。

摘　　要：为了解多种新型受体模型的适用性,利用正定矩阵分解/多元线性引擎2-物种比值(PMF/ME2-SR)、偏目标转换-正定矩阵分解(PTT-PMF)、正定矩阵分解(PMF)和化学质量平衡(CMB)这4种受体模型对我国北方典型城市细颗粒物(PM_(2.5))数据进行同步解析并互相验证.结果发现,燃煤源(25%~26%)、扬尘源(19%~21%)、二次硝酸盐(17%~19%)、二次硫酸盐(16%)、机动车源(13%~15%)、生物质燃烧源(4%~7%)和钢铁源(1%~2%)这7种主要污染源对研究地区PM_(2.5)有贡献.通过比较不同模型获得的源成分谱和源贡献以及计算各源的差异系数(CD)和平均绝对误差(AAE),发现4种模型的解析结果具有较高的一致性(平均CD值在0.6~0.7之间),但不同模型对各污染源中组分的识别存在差异.相比于传统PMF模型,PMF/ME2-SR模型由于纳入一次源类的特征比值,能够更好地区分源谱特征较为相似的源类,如扬尘源的CD和AAE分别比PMF模型低15%和54%;PTT-PMF模型以实测一次源谱和虚拟二次源谱为约束目标,计算的二次硫酸盐的CD和AAE分别为0.25和17%,比PMF低55%和23%,获得了更"纯净"的二次源类并识别了其他模型未识别的钢铁源,对源类的精细化解析更具优势.In order to understand the applicability of various new receptor models, four receptor models, including the positive matrix factorization/multilinear engine 2-species ratio(PMF/ME2-SR), partial target transformation-positive matrix factorization(PTT-PMF), positive matrix factorization(PMF), and chemical mass balance(CMB), were used to analyze and verify the atmospheric fine particulate matter(PM_(2.5)) data of a typical city in northern China. It was found that coal combustion(25%-26%), dust(19%-21%), secondary nitrate(17%-19%), secondary sulfate(16%), vehicle emissions(13%-15%), biomass burning(4%-7%), and steel(1%-2%) had a contribution to PM_(2.5). By comparing the source profiles and source contributions obtained by different models and calculating the coefficient of differences(CD) and average absolute error(AAE) of each source, we found that although the source apportionment results of the four models were in good agreement(the average CD value was between 0.6 and 0.7), there were still slight differences in the identification of some components in each source. Compared with the traditional model(PMF), the PMF/ME2-SR model can better identify sources with similar source profile characteristics, which is due to the component ratios of sources that are introduced. For example, the CD and AAE of dust sources were 15% and 54% lower than those of PMF, respectively. The PTT-PMF model takes the measured primary source profiles and virtual secondary source profiles as a constraint target, and the calculated CD and AAE of secondary sulfate were 0.25 and 17%, respectively, which were 55% and 23% lower than PMF. The PTT-PMF model can obtain more “pure” secondary sources and identify the pollution sources that are not identified by other models, which has more advantages in the refined identification of sources.

关 键 词：PM_(2.5)来源解析 正定矩阵分解/多元线性引擎2-物种比值模型(PMF/ME2-SR) 偏目标转换-正定矩阵分解模型(PTT-PMF) 新型受体模型 

分 类 号：X513[环境科学与工程—环境工程]