Source Attribution - Stage 4

01 Redefine objectives and set goals 

By Stage 4, you are familiar with the use of receptor modelling, including preparing input files, integrating different measurements, propagating errors, interpreting source apportionment results, and comparing them with model outcomes to understand existing uncertainties. The focus should now shift towards conducting advanced source apportionment using alternative approaches and modelling studies. This could involve using existing datasets, expanding the chemical database for organic tracers, incorporating other mass spectrometric datasets, and integrating modelling frameworks for interpretation and validation.

Continuous engagement with stakeholders and decision makers is also important to ensure that the findings can be directly used for policy interventions, or that changes can be made in the monitoring or source apportionment approach to account for this. For example, if source apportionment results show that a particular industrial activity is a major contributor to PM mass, this information can be shared with local authorities to highlight the issue. This could lead to stricter controls on emissions, changes in industrial practices, or improved handling of materials. To assess the impact of such actions, monitoring could be expanded, for instance by setting up temporary measurement sites near the industrial area or including additional pollutants targeting industrial activities (e.g. metals).  

02 Expand offline chemical database

At this stage, the focus should be on analysing primary organic molecular markers such as levoglucosan (indicative of biomass or wood burning), hopanes (associated with traffic emissions), and polycyclic aromatic hydrocarbons (PAHs, linked to combustion activities). Including these species in the dataset can significantly refine source contributions and provide a clearer understanding of combustion-related sources. In parallel, efforts should be made to develop updated source profiles for local emission sources using detailed chemical speciation data (organic + inorganic) (Hama et al., 2021; JRC, 2015; US EPA, 2024). This will enhance the accuracy and representativeness of source profiles for more reliable source attribution analysis. To obtain a source profile for your jurisdiction for a specific source, you need to conduct direct emission measurements to avoid any atmospheric processing. You can then analyse the collected samples for comprehensive chemical speciation, including both organic and inorganic species. This can be followed by the use of the source profile in receptor modelling (e.g. chemical mass balance) to refine source contributions.  

03 Use offline chemical data to identify sources

As outlined in Step 2, field samples are now being analysed for primary organic molecular markers. Including these markers, such as levoglucosan or hopanes, in receptor modelling can provide valuable insight into combustion-related sources and support more targeted decision-making for managing anthropogenic emissions (Srivastava et al., 2018). The selection of sources should follow the guidance provided in Stage 3 of the Source Attribution Guidance, and the resulting source profiles should be carefully validated using external tracers or diagnostic ratios (e.g. OC/levoglucosan or K/levoglucosan) to better assess contributions from sources such as biomass burning.

If necessary, constraints can be applied to improve model results. These can include using known relationships between species (e.g. specific ratios) or applying constraints to a particular species by “pulling down”. For example, levoglucosan would not be expected to contribute to dust-related sources, so a constraint can be applied to minimise its contribution to such sources and achieve physically meaningful results. Further guidance on applying and interpreting constraints can be found in the US EPA PMF manual (see pages 58–69) (US EPA, 2014). Source time series can also be combined with back-trajectory analysis to identify the geographical origins of pollution, providing additional evidence to support policy interventions. By examining the characteristics of identified sources, it is also possible to distinguish between primary and secondary contributions.  

The elemental carbon (EC) tracer method should also be considered as that has been widely used to estimate primary and secondary contributions, as it only requires EC and organic carbon (OC) as inputs. The key requirement for applying this method is to determine the primary OC to EC ratio, which can be obtained using several approaches (see resources below, Wu et al 2019). The information obtained on primary and secondary components can also be used to validate receptor modelling outputs, offering a practical and timely tool to inform emission control strategies.

In addition, analysing diagnostic ratios of polycyclic aromatic hydrocarbons (PAHs) provides a useful way to identify combustion sources. These ratios are based on the relative abundance of specific PAH compounds that are emitted differently depending on the source. For example, ratios such as retene to benzo(a)anthracene/(benzo(a)anthracene + chrysene) and indeno(123cd)pyrene/(indeno(123cd)pyrene + benzo(ghi)perylene) are capable of distinguishing between traffic and residential heating (Dvorska et al., 2011). While these ratios are not absolute indicators and can be influenced by atmospheric ageing and transport, they offer a practical first step for interpreting source contributions when combined with other measurements and modelling approaches. This is valuable for air quality management and policy-making, as it helps identify the main sources of pollution and supports more targeted and effective mitigation measures.  

04 Use black carbon data to investigate sources

Black carbon (BC) is a useful indicator of harmful particles from combustion sources (especially traffic) which may improve the assessment of policy actions to abate air pollution from combustion sources and understand their health impacts. By now you should have started to collect BC data for your jurisdiction. These data can be prepared for source apportionment, and validated using the guidelines provided by the Aerosol, Clouds and Trace Gases Research Infrastructure (ACTRIS) and Working Group WG 35 of the European Committee for Standardization (CEN). Aethalometer-based models can then be applied to separate contributions from different sources by using the wavelength dependence of light absorption by BC. In simple terms, emissions from fossil fuel combustion (e.g. traffic) and biomass burning absorb light differently across wavelengths, and this difference can be used to estimate their relative contributions (Savadkoohi et al. 2023). The model typically relies on assumptions about absorption Ångström exponents for different sources and provides time-resolved source contributions (e.g. traffic vs residential or commercial combustion). While the approach is relatively simple and can be applied routinely, care is needed in selecting appropriate parameters and interpreting results, especially in environments with mixed combustion sources.  Follow Savadkoohi et al., 2023 below to understand how this method can be used to provide useful, policy-relevant information on BC source contributions. 

05 Carry out source apportionment using low-cost sensor data 

By Stage 4, you may have access to a low-cost sensor (LCS) network that is running successfully (see Air Quality Monitoring Guidance, Stage 2, Step 4) and providing high-resolution spatial and temporal coverage of urban air quality. Low-cost sensors offer a practical way to capture variability that would not be feasible using only expensive regulatory or research-grade instruments, particularly in resource-limited settings. The next step is to prepare these data for further analysis, including validation against reference-grade measurements and appropriate handling of missing values. Once the data quality is ensured, statistical approaches such as k-means clustering, along with receptor modelling techniques as suggested in the studies below (Bousiotis et al., 2022 and Hagan et al., 2019) can be applied to infer potential sources.  

06 Use high time resolution instruments to investigate sources

At this stage, while not essential, a few major monitoring stations in your jurisdiction may begin measuring PM components using online mass spectrometry instruments (e.g. Aerosol Chemical Speciation Monitor – ACSM) (ARI, 2017). These measurements should be processed using established tools (e.g. the IGOR package) and validated against external markers to ensure data quality. Continuous measurements of key PM components such as organic matter, nitrate, sulphate, ammonium, and chloride can then be used to assess long-term trends and, when combined with back-trajectory analysis, help identify their geographical origins (Mandariya et al., 2024). Linking these high temporal resolution observations with meteorological parameters (e.g. ventilation coefficient) can further support the design of targeted intervention strategies (Gani et al., 2019). For example, high nitrate and ammonium levels can point towards agricultural emissions and traffic-related NOx, informing measures in the transport and farming sectors. High chloride levels in urban areas, where the impact of marine air is minimal, could be indicative of emissions from industrial or waste burning activities (Gunthe et al., 2021). Similarly, elevated organic matter, particularly during cold winter evenings, may indicate the influence of residential heating, supporting policies targeting the use of solid fuels for domestic purposes, including promoting cleaner alternatives for heating. High sulphate concentrations can suggest regional or industrial sources, highlighting the need for more coordinated emission control efforts at the regional level. In this way, PM component measurements help identify which sectors contribute most to pollution, allowing more targeted and effective air quality management actions. 

07 Identify emission sources during high pollution episodes

By this stage, a broad understanding of potential sources and their geographical origins has already been developed using various source attribution techniques. The focus can now shift towards identifying the drivers of short-term air quality exceedances and supporting timely interventions. These events are typically influenced by a combination of local emissions, regional transport, and meteorological conditions, making it necessary to use a combination of tools such as receptor modelling, regional air quality modelling, trajectory analysis, and continuous monitoring data. This may require strengthening partnerships with academic and research institutions that share the goal of improving current understanding of pollution levels and sources. The results such as source contributions to total PM mass and their origins can help decision-makers identify major sources and implement targeted actions (e.g. traffic restrictions or emission controls in the industrial sectors and in the residential sector for burning wood) during episodic events, as well as inform strategies to prevent future pollution episodes and design policies for inclusion in Graded Response Action Plans (GRAPs) (see Decision Support Guidance, Stage 4, Step 8). For example, Liu et al., 2019 applied a multi-model approach to investigate the drivers of haze formation in Beijing (see also examples from CAMS, 2024; Mandariya et al., 2024; Wang et al., 2025) 

08  Share source apportionment results with the public and decision makers 

Through the previous stages, you will have developed a good understanding of PM sources and how they vary over time across the monitored sites. At this stage, it becomes important to share this knowledge more widely with the public. This can include showing spatial patterns, such as regional-scale source maps from chemical transport models, to help people see how pollution changes across different areas. Clear and simple explanations of monitoring methods, along with key health information and visual illustrations, can make the information easier to understand for those without a technical background. Ongoing engagement with stakeholders and decision-makers will help ensure the platform remains useful and that the information being shared continues to support both public awareness and policy needs. You can refer to good practices outlined in the Public Engagement & Communications Guidance.