Integrating spatially distributed data into Positive Matrix Factorization to identify the hotspots of local emission sources

Abstract

The receptor model of Positive Matrix Factorization (PMF) is widely used to identify the air pollution sources of fine particulate matter (PM2.5). However, using data collected at a single site in PMF analysis limits its ability to characterize the source emission regions. To recognize the hotspots of emission sources, this study incorporated spatially distributed PM2.5 mass concentrations into PMF modeling to estimate the spatial distribution of source-specific PM2.5. A local emission source of road dust/civil construction was retrieved and selected for evaluation of the characterized hotspot regions. The spatial distribution of the corresponding source-specific PM2.5 contributions in the study area was compared with the land use features. The positive correlations (coefficients ≥0.40) were acquired between source-specific PM2.5 and land use characteristics, such as major road length and the number of construction sites. A leave-one-out cross-validation R2 of 0.48 was achieved using the land use regression model. These findings demonstrated the effectiveness of the proposed approach in capturing spatial variations of local emission sources. Future studies are suggested to expand the study area to further assess the extent of application of the PMF spatialization results.