Toward effective source apportionment using positive matrix factorization: experiments with simulated PM2.5 data.

Area of Science:

• Environmental Chemistry
• Atmospheric Science
• Chemical Engineering

Background:

• Particulate matter (PM2.5) impacts air quality and human health.
• Positive Matrix Factorization (PMF) is a receptor model used to identify sources of PM2.5.
• Understanding the relationship between PMF factors and actual emission sources is crucial for effective source apportionment.

Purpose of the Study:

• To clarify the link between PMF-resolved factors and real-world emission sources.
• To enhance PMF modeling strategies for PM2.5 source identification.
• To evaluate the interpretability of PMF factors using various statistical metrics.

Main Methods:

• Speciated PM2.5 data from a chemical transport model for two eastern US rural sites were analyzed using PMF.
• Goodness-of-fit metrics (chi2, R2) and root mean square difference were employed.
• Factor interpretability was assessed, with and without a priori knowledge and factor rotation.

Main Results:

• PMF factors frequently represent imperfect combinations of emission sources.
• An optimal number of factors is necessary to explain input data (e.g., R2 > 0.95); excess factors do not resolve minor sources without increased temporal resolution.
• Factor rotation guided by prior knowledge improves factor interpretability.

Conclusions:

• The number of factors in PMF models should be carefully selected to adequately explain data without overfitting.
• Incorporating external information, such as source markers or special events, enhances PMF analysis.
• Uncertainty weighting coefficients significantly affect PMF results, necessitating appropriate selection and validation, though data uncertainties can still impact solutions.