机器学习耦合受体模型量化驱动因素对PM2.5的影响效应
Machine learning coupled with receptor model quantifies the effect of driving factors onPM2.5
-
摘要: PM2.5受到污染源排放、大气物理化学以及气象条件等驱动因素的非线性影响. 随着在线观测技术的发展,利用大量观测数据量化实际环境下各驱动因素对PM2.5浓度的影响效应是研究的热点. 本文基于南开大学大气环境综合观测超级站2018年11月-2020年10月的在线观测数据分析了PM2.5的污染特征;利用PMF模型计算PM2.5的来源及其贡献,源解析结果显示观测期间二次源的贡献率最高,为44.7%,其他依次为燃煤源(23.6%),扬尘源(9.9%),生物质燃烧(7.2%),工业源的贡献率最小,为3.6%. 通过机器学习模型量化了一次源排放、大气氧化能力、气象条件等驱动因素对PM2.5的影响效应,结果表明,一次源排放对PM2.5影响效应的贡献 (54.2%)远高于其他驱动因素,气象条件对PM2.5影响效应的贡献次之 (32.6%),大气氧化能力对PM2.5影响效应的贡献较低 (13.2%). 本研究结果表明当前一次污染源排放仍是PM2.5污染最重要的成因,在大气污染防治工作中需要重点关注一次源的防控.Abstract: PM2.5 is nonlinearly affected by pollution source emission, atmospheric physicochemical, and meteorological conditions. With the development of online observation technology, quantifing the effect of driving factors on PM2.5 in actual atmosphere through observation datas becomes a research hotspot. Based on the online observation data of the atmospheric environment comprehensive observation superstation of Nankai University from November 2018 to October 2020, this paper analyzes the pollution characteristics of PM2.5; and uses the PMF model to identify the source and contribution of PM2.5, the results show that the contribution of secondary sources during the observation period is the highest (44.7%), followed by coal-fired sources (23.6%), dust sources (9.9%), biomass combustion (7.2%), and the contribution of industrial sources is the lowest (3.6%); This paper also explores the effects of drivers such as emission sources, atmospheric oxidation capacity, and meteorological conditions on PM2.5 through machine learning model. The results show that the effect of emission sources (54.2%) is higher than other drivers, followed by the effect of meteorological conditions (32.6%), and the effect of atmospheric oxidation capacity (13.2%) is lower. At present, primary source emission is still the most important cause of PM2.5 pollution, and it is necessary to focus on the treatment of primary sources in the prevention and control of air pollution.
-
Key words:
- receptor model /
- random forest /
- pollution characteristics of PM2.5 /
- driving factor
点击查看大图
计量
- 文章访问数: 178
- HTML全文浏览量: 29
- PDF下载量: 114
- 被引次数: 0