1,1,1,2,3- 1College of Environmental Science and Engineering, Nankai University, Tianjin, China
- 2Department of Public Health Sciences, University of Rochester School of Medicine and Dentistry, Rochester, NY, USA
- 3Institute for a Sustainable Environment, Clarkson University, Potsdam, NY, USA
Reliable source apportionment of ambient air pollutants is essential for effective air quality management. Positive Matrix Factorization (PMF) has been the most widely applied receptor-based method for ambient particulate matter (PM) (Hopke et al., 2020). Although EPA PMF v5 incorporates approaches to evaluate uncertainties in source profiles due to measurement error and rotational ambiguity (Paatero et al., 2014), it does not provide quantitative estimates of uncertainties in source contributions. Previous attempts to address this issue have been limited to sensitivity tests rather than rigorous uncertainty analyses. Here we introduce a novel approach to quantify uncertainties in source contributions (G matrix) within PMF solutions. By combining PMF-derived factor profiles (F matrix) with observed concentration data, we employ an Effective Variance Least Squares (EVLS) reverse-calculation framework to estimate the standard deviation of each source contribution, yielding a more rigorous and quantitative assessment of PMF uncertainties. A total of 837 valid daily filter-based speciation samples, collected from May 2013 to February 2019 in Tianjin, China (Dai et al., 2023), were used for PMF modeling and methodological testing. Compared with conventional PMF analysis, the PMF-EVLS approach yields both source contribution concentrations and their associated standard deviations. These uncertainties, expressed as standard deviations, reflect the combined influence of various error sources (e.g., model assumptions, measurement errors). The proposed PMF–EVLS method was demonstrated to effectively estimate the uncertainty of source-specific PM2.5 concentrations, thereby enhancing the reliability of source-specific health effect assessments and supporting air quality management decisions.
Refs:
1. Dai, Q., Chen, J., Wang, X., Tian, Y., et al. (2023). Environ. Pollut. 325:121344.
2. Hopke, P.K., Dai, Q., Li L., Feng Y. (2020). Sci. Total Environ. 740, 140091.
3. Paatero, P., Eberly, S., Brown, S.G., Norris, G.A. (2014). Atmos. Meas. Tech. 7, 781–797.
How to cite: Dai, Q., Chen, J., Zhang, X., Tian, Y., Feng, Y., and Hopke, P.: PMF-resolved source contribution uncertainty estimation via effective variance least squares, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10765, https://doi.org/10.5194/egusphere-egu26-10765, 2026.
