Decomposing PM_2.5 concentrations in urban environments into meaningful factors: 1. Separating the contribution of local anthropogenic activities from background and long-range transport

Fine particulate matter (PM_2.5) is a complex mixture of aerosol particles with varying properties and sources, both local and distant. In areas lacking detailed monitoring of PM_2.5 speciation, the common source-apportionment analyses are not applicable. This study demonstrates an alternative framework for estimating sources and processes that affect observed PM_2.5 concentrations when information on the particle composition is unavailable. Eight years (2012–2019) of half-hourly PM_2.5 observations from 10 air quality monitoring (AQM) stations, clustered according to their airmass transport sector were analyzed, using Non-negative Matrix Factorization (NMF). Factors were determined based on their variation in time, space, and between airmass sectors. Employing a supervised machine-learning model provided insights into the relationships between the extracted factors, meteorological parameters and co-measured airborne pollutants. Factor interpretations were evaluated through comparisons with measurements of PM_2.5 species from a nearby Surface PARTiculate mAtter Network (SPARTAN) station. The NMF successfully separated background factors from an urban anthropogenic-activity factor, with the latter accounting for approximately 60 % of the observed PM_2.5 levels in Tel Aviv (∼10±6μg/m³). Positive monotonic relationships were observed between the PM_2.5 urban anthropogenic-activity factor and measurements of nitrogen oxides (NO_x) and absolute humidity (AH), representing the impact of traffic emissions and hygroscopic growth, respectively. The summer background factor was found to represent long-range transport (LRT) from Europe, showing a good agreement (R² = 0.81) with ammonium sulphate concentrations. Our results demonstrate that a spatial NMF analysis can reliably estimate contributions of different sources with distinct compositions and properties to the total observed PM_2.5. Using such an analysis, future environmental health studies could assess health risks associated with exposure to distinct PM_2.5 fractions. This information may assist decision makers to set environmental targets for abating PM_2.5 with specific compositions and properties.