PMF (Positive Matrix Factorization) 수용모델을 활용한 오염원 기여도 분석 고도화 연구

• Published in: 한국대기환경학회지(국문) Vol. 38; no. 4; pp. 493 - 507
• Main Authors: 유일한(Ilhan Ryoo), 박지은(Jieun Park), 김태연(Taeyeon Kim), 류지원(Jiwon Ryu), 정연승(Yeonseung Cheong), 안준영(Joonyoung Ahn), 이승묵(Seung-Muk Yi)
• Format: Journal Article
• Language: Korean
• Published: 한국대기환경학회 01.08.2022
• Subjects: 대기과학; PMF; Dispersion Normalized PMF; Source apportionment; Moving Window PMF
• ISSN: 1598-7132; 2383-5346
• DOI: 10.5572/KOSAE.2022.38.4.493

The positive matrix factorization (PMF) receptor model tracks sources of fine particle (PM2.5) based on data on the concentration and uncertainty of PM2.5 in the atmosphere, and is widely used in establishing air quality management policies worldwide. However, the conventional PMF model does not take meteorological variables such as mixing layer height (MLH) and wind speed into account which can affect the contribution of pollutants. Also, it is difficult to distinguish events over a specific period of time because one input data is constructed for the entire model period, and the uncertainty of model results cannot be calculated. Therefore, this study aimed to supplement the limitations of the conventional model by applying two advanced methods: DN-PMF and Moving Window PMF based on real-time data from the Seoul Metropolitan Air Environment Research Institute in 2019 and 2020. The DN-PMF model results showed lower contribution concentration of sources such as industry, oil combustion, aged sea salt, and mobile compared to that of the PMF, which means that the ventilation coefficient was low. For the Moving Window PMF, all six SETs showed similar trends, and an additional fireworks source was separated from SETs 5 and 6 including a specific period. In addition, with the addition of fireworks source, the standard deviation between sources, excluding sources whose contribution concentration changed, was small within 1.0. If the advantages of these two methods are appropriately utilized, it can be used as a basis for establishing air quality policies more effectively in the future. KCI Citation Count: 0