Improving apportionment of PM2.5 using multisite PMF by constraining G-values with a priori information
Qili Dai1, Philip K Hopke2, Xiaohui Bi1
1State 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.
The Science of the Total Environment
|June 6, 2020
Summary
Constraining positive matrix factorization (PMF) with wind data improves source apportionment accuracy. This method enhances the reliability of particulate matter (PM2.5) source identification, especially near industrial emissions.
Area of Science:
- Environmental Science
- Atmospheric Chemistry
- Data Analysis
Background:
- Factor analysis methods like positive matrix factorization (PMF) suffer from rotational ambiguity, hindering accurate source apportionment.
- Prior constraints on PMF solutions primarily focused on source profiles or simultaneous PM2.5/PM10 data, with limited exploration of source contribution constraints.
Purpose of the Study:
- To develop and evaluate a methodology for improving PMF accuracy by constraining source contributions using local wind information.
- To apply this methodology to multisite PMF analysis for better identification of local and regional pollution sources.
Main Methods:
- Combined individual PMF analyses from three monitoring sites (INDUS, URBAN, RURAL) into a multisite PMF analysis.
- Constrained source contributions using local wind direction data to reduce rotational ambiguity.
- Utilized coefficient of divergence and Pearson correlation analysis to identify local vs. regional source contributions.
- Quantitatively estimated source contributions using the Lenschow approach.
Main Results:
- Identified seven common factors: nitrate (28.7%), sulfate (22.5%), coal combustion (19.3%), road traffic (12.8%), biomass burning (6.4%), soil (5.4%), and metallurgical industry (4.9%).
- Constrained solutions significantly improved results compared to the base run by reducing rotational space.
- Determined average local source contributions of 52.4% (INDUS) and 47.7% (URBAN).
- Metallurgical industry was identified as a major local source, while sulfate was predominantly regional.
Conclusions:
- Constraining PMF with wind direction data is an effective method for improving the accuracy and reliability of source apportionment, particularly in areas with point source emissions.
- This approach enhances the ability to distinguish between local and regional pollution sources.
- The methodology provides more robust solutions for multisite PMF analyses.
