Real-Time Source Apportionment at an Urban Traffic Site in Athens during the MI-TRAP 2025 Campaign

Particulate matter (PM) pollution at urban traffic sites reflects a complex mixture of exhaust and non-exhaust traffic emissions together with contributions from other urban sources such as biomass burning, cooking, secondary aerosol formation, and natural inputs. Understanding the temporal variability and seasonal evolution of these sources is essential for designing effective mitigation strategies, yet it is often constrained by traditional offline source apportionment methods. Real-time source apportionment (RT-SA) offers the ability to continuously resolve PM sources and track changes in their chemical composition at high time resolution.This study presents results from the MI-TRAP 2025 measurement campaign conducted at a single urban traffic site in Athens, Greece, within the framework of the EU Horizon Europe MI-TRAP project. The site was strongly influenced by road traffic emissions while simultaneously capturing the full spectrum of PM sources typically present in an urban traffic environment, including residential biomass burning, cooking activities, secondary aerosol, and natural sources.An integrated ACSM–Xact–Aethalometer (AXA) system was deployed in combination with the SoFi RT software to perform real-time PM source apportionment. The system provides simultaneous measurements of organic aerosol composition, elemental concentrations, and black carbon, enabling detailed characterization of both primary and secondary PM sources. Traffic-related emissions dominated the PM mass, while secondary components accounted for a significant percentage of the total PM.A rolling-window RT-SA approach was applied to capture temporal and seasonal changes in source profiles and contributions. This analysis revealed distinct seasonal variability in the chemical composition of several sources, particularly in biomass burning. Changes in elemental markers and organic aerosol signatures reflected shifts in fuel use, atmospheric processing, and driving conditions between seasons. The rolling-window methodology proved essential for resolving these evolving source characteristics, which would be obscured in a single static source apportionment model.The SoFi RT framework enabled continuous and near-instantaneous source apportionment with automated data processing. Comparison between real-time results and an optimized offline approach showed good agreement, confirming the robustness of the real-time methodology. Overall, the MI-TRAP Athens campaign demonstrates the capability of real-time source apportionment combined with rolling-window analysis to provide new insights into the seasonal dynamics and chemical evolution of PM sources at urban traffic sites.