Machine learning integrated PMF model reveals influencing factors of ozone pollution in a coal chemical industry city at the Jiangsu-Shandong-Henan-Anhui boundary

IF 4.2 2区环境科学与生态学 Q2 ENVIRONMENTAL SCIENCES

Atmospheric Environment Pub Date : 2024-11-02 DOI:10.1016/j.atmosenv.2024.120916

Chaolong Wang , Xiaofei Qin , Yisheng Zhang , Dantong Liu , Wenxin Tao , Ming Wang , Sufan Zhang , Jianli Yang , Jinhua Du , Shanshan Cui , Dasa Gu , Yingjie Sun , Chenying Lv

{"title":"Machine learning integrated PMF model reveals influencing factors of ozone pollution in a coal chemical industry city at the Jiangsu-Shandong-Henan-Anhui boundary","authors":"Chaolong Wang , Xiaofei Qin , Yisheng Zhang , Dantong Liu , Wenxin Tao , Ming Wang , Sufan Zhang , Jianli Yang , Jinhua Du , Shanshan Cui , Dasa Gu , Yingjie Sun , Chenying Lv","doi":"10.1016/j.atmosenv.2024.120916","DOIUrl":null,"url":null,"abstract":"<div><div>Zaozhuang, located at the center of the boundary between Jiangsu, Shandong, Henan, and Anhui, contains coal and heavy industries. Zaozhuang has experienced severe O3 pollution in recent years and it is crucial to identify the key drivers. This study aims to deeply excavate and analyze the formation mechanism of O3 in Zaozhuang based on hourly measured volatile organic compound (VOC) concentration data for the year 2023, combined with meteorological factors and other atmospheric pollutants, using a machine learning model in combination with the SHapley Additive Properties Interpretation method and Positive Matrix Factorization model. The results show the important contributions of meteorological factors to O3 production, especially solar radiation and temperature. Among atmospheric pollutants, VOCs are the main contributors, with significant effects from alkene and oxygenated VOCs, whereas propene and acetone have the most critical individual impacts on local O3 production. O3 peaked in June and August, with June seeing added contributions from temperature, and a higher chemical variable contribution than meteorological factors in August, led by NO2, OVOCs, and alkenes. The effects of the six emission sources on O3 formation in Zaozhuang showed that chemical emission sources (5.98 μg/m3), combustion sources (3.75 μg/m3), and solvent use sources (3.06 μg/m3) were the main drivers. The solvent source exhibited the most significant change on the O3 polluted day, with a relative increase of 115%. This relative increase was significantly higher than that of the other sources. During persistent pollution events with the highest levels of O3, the use of solvents made the greatest contribution to the emission sources, representing 23% of the total impact of the emission sources. Therefore, an integrated approach using machine learning, SHapley Additive Properties Interpretation, and Positive Matrix Factorization rapidly diagnoses the causes of O3 pollution at different timescales and provides a basis for targeted control measures.</div></div>","PeriodicalId":250,"journal":{"name":"Atmospheric Environment","volume":"342 ","pages":"Article 120916"},"PeriodicalIF":4.2000,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Atmospheric Environment","FirstCategoryId":"93","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1352231024005910","RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENVIRONMENTAL SCIENCES","Score":null,"Total":0}

引用次数: 0

Abstract

Zaozhuang, located at the center of the boundary between Jiangsu, Shandong, Henan, and Anhui, contains coal and heavy industries. Zaozhuang has experienced severe O₃ pollution in recent years and it is crucial to identify the key drivers. This study aims to deeply excavate and analyze the formation mechanism of O₃ in Zaozhuang based on hourly measured volatile organic compound (VOC) concentration data for the year 2023, combined with meteorological factors and other atmospheric pollutants, using a machine learning model in combination with the SHapley Additive Properties Interpretation method and Positive Matrix Factorization model. The results show the important contributions of meteorological factors to O₃ production, especially solar radiation and temperature. Among atmospheric pollutants, VOCs are the main contributors, with significant effects from alkene and oxygenated VOCs, whereas propene and acetone have the most critical individual impacts on local O₃ production. O₃ peaked in June and August, with June seeing added contributions from temperature, and a higher chemical variable contribution than meteorological factors in August, led by NO₂, OVOCs, and alkenes. The effects of the six emission sources on O₃ formation in Zaozhuang showed that chemical emission sources (5.98 μg/m³), combustion sources (3.75 μg/m³), and solvent use sources (3.06 μg/m³) were the main drivers. The solvent source exhibited the most significant change on the O₃ polluted day, with a relative increase of 115%. This relative increase was significantly higher than that of the other sources. During persistent pollution events with the highest levels of O₃, the use of solvents made the greatest contribution to the emission sources, representing 23% of the total impact of the emission sources. Therefore, an integrated approach using machine learning, SHapley Additive Properties Interpretation, and Positive Matrix Factorization rapidly diagnoses the causes of O₃ pollution at different timescales and provides a basis for targeted control measures.

Abstract Image

查看原文本刊更多论文

机器学习集成 PMF 模型揭示江苏-山东-河南-安徽交界煤化工城市臭氧污染的影响因素

枣庄位于江苏、山东、河南和安徽交界处的中心，煤炭和重工业发达。近年来，枣庄经历了严重的臭氧污染，找出其关键驱动因素至关重要。本研究旨在基于 2023 年每小时实测的挥发性有机化合物（VOC）浓度数据，结合气象因子和其他大气污染物，利用机器学习模型，结合 SHapley Additive Properties Interpretation 方法和正矩阵因子化模型，深入挖掘和分析枣庄市 O3 的形成机理。结果表明，气象因素对 O3 的产生有重要影响，尤其是太阳辐射和温度。在大气污染物中，挥发性有机化合物是主要的贡献者，其中烯烃和含氧挥发性有机化合物的影响显著，而丙烯和丙酮对当地 O3 生成的影响最为关键。O3 在 6 月和 8 月达到峰值，其中 6 月温度的贡献更大，8 月化学变量的贡献高于气象因素，主要是二氧化氮、OVOC 和烯烃。六种排放源对枣庄臭氧形成的影响表明，化学排放源（5.98 μg/m3）、燃烧源（3.75 μg/m3）和溶剂使用源（3.06 μg/m3）是主要的驱动因素。在 O3 污染日，溶剂源的变化最为显著，相对增加了 115%。这一相对增幅明显高于其他污染源。在 O3 水平最高的持续污染事件中，溶剂的使用对排放源的贡献最大，占排放源总影响的 23%。因此，利用机器学习、SHapley Additive Properties Interpretation 和正矩阵因式分解的综合方法可快速诊断不同时间尺度的臭氧污染成因，并为采取有针对性的控制措施提供依据。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Atmospheric Environment 环境科学-环境科学

CiteScore

9.40

自引率

8.00%

发文量

458

审稿时长

53 days

期刊介绍： Atmospheric Environment has an open access mirror journal Atmospheric Environment: X, sharing the same aims and scope, editorial team, submission system and rigorous peer review. Atmospheric Environment is the international journal for scientists in different disciplines related to atmospheric composition and its impacts. The journal publishes scientific articles with atmospheric relevance of emissions and depositions of gaseous and particulate compounds, chemical processes and physical effects in the atmosphere, as well as impacts of the changing atmospheric composition on human health, air quality, climate change, and ecosystems.