A fatoração PMF - Positive Matrix Factorization - é um método de resolução de problemas em que variáveis observadas conjuntamente são modeladas como a combinação linear de fatores potenciais, representada pela multiplicação de duas matrizes. Este método tem sido utilizado principalmente em áreas de estudo em que as variáveis observadas são sempre não negativas, e quando é aplicada uma modelagem fatorial ao problema. Assume-se a premissa de que são resultantes da multiplicação de duas matrizes com entradas não negativas, ou seja, os fatores potenciais, e os poderadores da combinação linear são desconhecidos, e têm valores não negativos. Neste método além da possibilidade de restringir a busca dos valores das matrizes da fatoração a valores não negativos, também é possível incluir a medida de incerteza relacionada a cada observação no processo de obtenção da fatoração como um modo de reduzir o efeito indesejado que valores outliers podem causar no resultado. Neste trabalho é feito um estudo de sensibilidade da fatoração PMF em relação a algumas medidas de incertezas presentes na literatura, utilizando o software EPA PMF 5.0 com ME-2. Para realizar este estudo foi desenvolvida uma metodologia de simulação de base de dados a partir de perfis de fontes emissoras conhecidas incluindo valores outliers, e aplicação sequencial da fatoração PMF com o software ME-2 nas bases de dados simuladas. / The PMF factorization - Positive Matrix Factorization - is a problem solving method in which jointly observed variables are modeled as a linear combination of potential factors, represented by the multiplication of two matrices. This method has been used primarily in study areas in which the observed variables are always non negative, and when it is applied a factor modeling in the problem. It is made the assumption that the variables in study come from the two matrices multiplication both having non negative components, i.e., the potential factors, and the linear combination values are unknown, and all of them have non negative values. In this method, besides the possibility of constraining the search of the matrix factorization values on non negative values, it is also possible to include the uncertainty measure related to each observation on factorization process as a way to reduce the undesired effect which outliers can cause to the outcome. This paper presents a study of the sensitivity of the factorization PMF over some uncertainties measures present in literature, using EMP PMF 5.0 with ME-2 software. To carry out this study was developed a methodology of database simulation from known emitting sources profiles including outliers values, and a sequential application of PMF factorization with ME-2 software in simulated databases.

A fatoração PMF - Positive Matrix Factorization - é um método de resolução de problemas em que variáveis observadas conjuntamente são modeladas como a combinação linear de fatores potenciais, representada pela multiplicação de duas matrizes. Este método tem sido utilizado principalmente em áreas de estudo em que as variáveis observadas são sempre não negativas, e quando é aplicada uma modelagem fatorial ao problema. Assume-se a premissa de que são resultantes da multiplicação de duas matrizes com entradas não negativas, ou seja, os fatores potenciais, e os poderadores da combinação linear são desconhecidos, e têm valores não negativos. Neste método além da possibilidade de restringir a busca dos valores das matrizes da fatoração a valores não negativos, também é possível incluir a medida de incerteza relacionada a cada observação no processo de obtenção da fatoração como um modo de reduzir o efeito indesejado que valores outliers podem causar no resultado. Neste trabalho é feito um estudo de sensibilidade da fatoração PMF em relação a algumas medidas de incertezas presentes na literatura, utilizando o software EPA PMF 5.0 com ME-2. Para realizar este estudo foi desenvolvida uma metodologia de simulação de base de dados a partir de perfis de fontes emissoras conhecidas incluindo valores outliers, e aplicação sequencial da fatoração PMF com o software ME-2 nas bases de dados simuladas. / The PMF factorization - Positive Matrix Factorization - is a problem solving method in which jointly observed variables are modeled as a linear combination of potential factors, represented by the multiplication of two matrices. This method has been used primarily in study areas in which the observed variables are always non negative, and when it is applied a factor modeling in the problem. It is made the assumption that the variables in study come from the two matrices multiplication both having non negative components, i.e., the potential factors, and the linear combination values are unknown, and all of them have non negative values. In this method, besides the possibility of constraining the search of the matrix factorization values on non negative values, it is also possible to include the uncertainty measure related to each observation on factorization process as a way to reduce the undesired effect which outliers can cause to the outcome. This paper presents a study of the sensitivity of the factorization PMF over some uncertainties measures present in literature, using EMP PMF 5.0 with ME-2 software. To carry out this study was developed a methodology of database simulation from known emitting sources profiles including outliers values, and a sequential application of PMF factorization with ME-2 software in simulated databases.

La connaissance des principales sources de l’aérosol permet d’améliorer, d’adapter et de cibler les mesures prises pour réduire les concentrations de particules fines. Ainsi, l’identification et la hiérarchisation des sources de particules fines sont des étapes essentielles à la mise en place d'une politique efficace d'amélioration de la qualité de l'air. Le travail mené durant cette thèse s’inscrit dans cette démarche puisqu'il avait pour objectif de quantifier les sources de PM2.5 en milieu industriel. Afin de répondre à cet objectif, deux campagnes de prélèvements ont été réalisés dont une sous les vents des principales activités industrielles afin de caractériser leurs émissions (profils) et une en zones urbaines caractéristiques de l’exposition de la population aux particules fines. Les résultats ont permis d'obtenir des empreintes représentatives des principales activités industrielles de la zone d'étude. L’analyse ME-2 menée a permis, avec la combinaison d’analyses radiocarbones, de déterminer que la source de combustion de biomasse est la source majoritaire pendant l’automne et l’hiver où les épisodes de PM2.5 ont été observés. La source industrielle est la source majoritaire des PM2.5 au printemps et en été mais ne constitue pas un driver fort de la concentration des PM2.5. Toutefois, cette étude a montré que les sources industrielles impactent significativement la population de particules (taille, composition, etc.) dans la zone d’étude. / In order to limit the impact of air quality on human health, public authorities need reliable and accurate information on the sources contribution. So, the identification of the main sources of PM2.5 is the first step to adopt efficient mitigation policies. This work carry out in this thesis take place in this issue and was to determine the main sources of PM2.5 inside an industrial area. To determinate the main sources of PM2.5, two campaigns were lead to collect daily PM2.5 to: 1/ determine the enrichment of atmospheric pollutants downwind from the main industrial activities and 2/ collect PM2.5 in urban areas characteristic of the population exposition. Results allowed to obtain very representative profiles for the main industrial activities implanted inside the studied area. ME-2 analysis, combined to radiocarbon measurements, allowed to highlight the very high impact of Biomass Burning sources for all the PM2.5 pollution events recorded from early autumn to March. This study showed that industrial sources, even if they are the major sources during spring and summer, are not the major PM2.5 driver. However, this study highlights that industrial sources impact significantly the aerosol population (size, composition, etc.) in the studied area.

Atmospheric aerosols have been shown to have a significant impact on air quality and health in urban environments. Organic aerosols (OA) are one of the main constituents of submicron particulate matter. They are composed of thousands of different chemical species, which makes it challenging to identify and quantify their sources. OA sources have been previously studied; however quantitative knowledge of aerosol composition and their processes in urban environments is still limited. The results presented here investigate OA, their chemical composition and sources as well as their interaction with gases. On-line measurements of species in the particle and the gas phase were performed both from field-based and laboratory studies. Aerosol Mass Spectrometers (AMS) were used together with the Chemical Ionisation Mass Spectrometer (CIMS) and the Filter Inlet for Gases and AEROsols (FIGAERO). Two ambient datasets were analysed to develop methods for source apportionment, using the Multilinear Engine (ME-2), in order to gain new insights into aerosol sources in Manchester and London. Long-term measurements in London allowed the opportunity to perform seasonal analysis of OA sources and look into the relationship of hydrogen-like OA (HOA) and heavy- and light-duty diesel emissions. The seasonal analysis provided information about OA sources that was not possible to observe on the long-term analysis. During Bonfire Night in Manchester, with high aerosol concentrations, particularly biomass burning OA (BBOA), it was possible to identify particulate organic oxides of nitrogen (PON), with further identification of primary and secondary PON and their light absorbing properties. Through laboratory work, new insights into cooking organic aerosols (COA) were gained, a higher relative ion efficiency (RIEOA) value of around 3.3 for OA-AMS compared with the typical RIEOA of 1.4 was determined, which implies COA concentrations are overestimated when using the RIEOA value of 1.4. Dilution showed to have a significant effect on food cooking experiments, increasing both the gas/particle ratios and the O:C ratios. The data generated in this work, OA-AMS mass spectra and markers from both gas and particle phase identified with FIGAERO-CIMS, provide significant information that will contribute to the improvement of source apportionment in future studies. This work investigates OA, with a focus on primary organic aerosols originated from anthropogenic activities. These scientific findings increase our understanding of OA sources and can help to improve inventories and models as well as to develop plans and policies to mitigate the air pollution in urban environments.

Les particules fines (PM2.5) suscitent de plus en plus l’intérêt des pouvoirs publics en raison de leurs effets néfastes sur la qualité de l’air et la santé humaine. La mise en place des politiques de réductions efficaces des émissions requière une connaissance détaillée des contributions des principales sources aux concentrations ambiantes en PM. Ainsi, cette thèse a pour objectifs de caractériser la composition chimique des PM2.5, et de quantifier leurs sources d’émissions à Marseille. Pour se faire, une campagne de mesure d’un an (2011-2012) a été conduite sur le site de fond urbain de « Cinq avenues ». La spéciation chimique complète des filtres collectés a été réalisée, et 3 modèles-récepteurs ont été utilisés: CMB (Chemical Mass Balance), PMF (Positive Matrix Factorization), et ME-2 (Multilinear Engine). Bien que basés sur des concepts sensiblement différents, l’exercice d’intercomparaison des sorties de ces modèles a montré globalement un bon accord pour l’estimation des contributions de la combustion de biomasse (entre 23 et 33% en moyenne annuelle) et du trafic véhiculaire (14-26%). En revanche, des différences significatives sont observées pour la source industrielle (1-18%) et le sulfate d’ammonium (12-30%). Cette étude a mis en évidence une contribution importante de la matière organique (OM) qui représente en moyenne 42% des PM2.5. Quant à la quantification des sources, l’un des résultats marquants est la mise en évidence de deux types d’aérosols de combustion de biomasse, dont l’un provient très probablement du brûlage à l’air libre de déchets verts. Ce dernier peut même être considéré comme un contributeur majeur des PM2.5 à l’automne et en début d’hiver. / Fine particulate matter (PM2.5) has received considerable attention due to its impact on human health and air quality. Therefore, effective plans for human health protection require a detailed knowledge of the most relevant PM emission sources and their contributions to the ambient PM levels. Thus, this thesis aims to characterize the chemical composition of PM2.5 collected in Marseille area, and quantify the impacts of the main emission sources. To meet these objectives, a one-year monitoring campaign was conducted at the urban background site of “Cinq avenues” during the period of 2011-2012. A detailed chemical characterization of the collected PM2.5 filters was performed, and 3 receptor models were applied to this database: CMB (Chemical Mass Balance), PMF (Positive Matrix Factorization), and ME-2 (Multilinear Engine). Although based on significantly different concepts, the intercomparison exercise of the output data of the used models has generally showed a good agreement in estimating the source contributions of biomass burning (representing between 23 and 33% on annual average) and vehicular traffic (between 14 and 26%). In contrast, significant differences were observed for the industrial (1-18%) and ammonium sulfate (12-30%) sources. This study highlighted the significant contribution of organic matter (OM), which represents 42% of the PM2.5 mass, on average. Regarding the source apportionment results, one of the most striking findings is the identification of two types of biomass burning aerosol, one of which probably comes from open burning of green waste. The latter can even be considered a major contributor to the PM2.5 mass during fall and early winter

Based on the concentrations of ten heavy metals (As, Cd, Cr, Cu, Hg, Mn, Ni, Pb, Zn, Fe) in 144 road dust samples collected from 36 sites across 4 seasons from 2016 to 2017 in Beijing, this study systematically analyzed the levels and main sources of health risks in terms of their temporal and spatial variations. A combination of receptor models (positive matrix factorization and multilinear engine-2), human health risk assessment models, and Monte Carlo simulations were used to apportion the seasonal variation of the health risks associated with these heavy metals. While non-carcinogenic risks were generally acceptable, Cr and Ni induced cautionary carcinogenic risks (CR) to children (confidence levels was approximately 80% and 95%, respectively). Additionally, fuel combustion posed cautionary CR to children in all seasons, while the level of CR from other sources varied, depending on the seasons. Heavy metal concentrations were the most influential variables for uncertainties, followed by ingestion rate and skin adherence factor. The values and spatial patterns of health risks were influenced by the spatial pattern of risks from each source.