Implementation of real-time source apportionment approaches using the ACSM–Xact–Aethalometer (AXA) setup with SoFi RT: the Athens case study

Abstract

Air pollution, particularly from particulate matter (PM), poses serious public health and environmental risks, especially in urban areas. To address this, accurate source apportionment (SA) of PM is essential for effective air quality management. Traditional SA approaches often rely on offline data collection, limiting timely responses to pollution events. SA applied on data from online techniques, especially with high temporal resolution, is advantageous over offline techniques, enabling the study of the diurnal variability of emission sources and also the study of specific events. Recent technological advancements now enable real-time SA, allowing continuous, detailed analysis of pollution sources. This study presents the first application of the ACSM–Xact–Aethalometer (AXA) setup integrated with SoFi RT software for real-time source apportionment of PM in Athens, Greece. The AXA setup integrates chemical, elemental, and black carbon (BC) data streams, covering a broad spectrum of PM components and capturing a comprehensive representation of PM sources in an urban environment. SoFi RT handles data from the AXA instruments as separate inputs within a single matrix, placing them in distinct diagonal blocks. Each main instrument's data (ACSM, Xact) is processed independently, with the model applying instrument-specific constraints and generating separate source factors, effectively performing two parallel source apportionments in a single run of the ME-2 solver. Equivalent sources identified across the two instruments are then combined post-analysis to provide a unified interpretation of source contributions. The apportionment of BC to BCsf and BClf (solid fuel and liquid fuel) can be performed in either of the main instrument experiments and does not require dedicated processing. The results demonstrate that traffic-related emissions are the largest contributors to PM, with significant contributions from secondary species such as sulfate, nitrate, ammonium, and secondary organic aerosols, which together accounted for approximately 57 % of the PM mass. Primary sources such as biomass burning and cooking contributed around 10 % each, with natural sources like dust and sea salt comprising the remainder. The SoFi RT software is employed for continuous SA, offering automated analysis of PM sources in near real time (minutes after the measurements). Our findings demonstrate that this setup effectively identifies major pollution sources. This work underscores the AXA system's potential for advancing urban air quality monitoring and informs targeted interventions to reduce PM pollution.