Arsenic (As) exposure is a cause of growing global concern. Exposure to As leads to various problems and can be fatal as well. Exposure to arsenic can be due to contact with various components of the environment that are contaminated by As. To remediate the effects of As, it is important to understand how this exposure occurs, and which parameters need to be taken into account for remediation. To do this, it is important to understand the extent and nature of exposure. This can be done through exposure modelling.The aim of this Master thesis is to develop an exposure model to perform risk assessment of As exposure in agricultural lands of Europe. The exposure model was developed in collaboration with KTH Royal Institute of Technology and Kemakta Konsult AB. The research is part of the European Union project AgriAs, ‘Evaluation and Management of Arsenic Contamination in Agricultural Soil and Water’ with six participating organizations from Finland, Germany, France and Sweden. Two target sites were chosen to perform a risk assessment using the developed model. The target sites were Freiberg, Saxony, Germany and Verdun, France. This is a preliminary model to practice the methodology. The final AgriAs model will be compiled together with the AgriAs Consortium and published jointly by all partners involved.The model was developed based on the identified exposure pathways of human exposure to As in agricultural regions, the different types of site-input data, and model-specific parameters. The identified exposure pathways are inhalation of dust, intake of As contaminated food stuff, ingestion of As contaminated water, accidental ingestion of soil, and dermal contact. The exposure model was developed as an equilibrium model, and the pathway equations were based on site-concentration and equilibrium transfer factors. The model was tested and validated by comparing with the Swedish Guideline model. The model was tested for sensitivity to the input parameter values as well. The validated model was then used to perform a Risk Assessment on the two target sites of Germany and France. Recommendations on further development of the model, and data collection required to make the model produce more site-specific results were also given.Keywords: Arsenic, Exposure model, Model development, Risk assessment, Agricultural lands, EuropeNote: Some parts have been removed from the public version of this master thesis until 2019, and the full research will be published by AgriAs in 2019. / AgriAs

La démarche d'évaluation des risques sanitaires inclue dans ses étapes l'évaluation de l'exposition professionnelle. S'agissant de l'exposition par inhalation, différents modèles sont disponibles pour caractériser ou estimer les niveaux d'exposition. Les objectifs de cette thèse sont d'étudier et comparer les principales démarches existantes, voire d'en développer de nouvelles, afin de participer à leur validation et de définir leur places respectives en complément de la métrologie quantitative. Un premier chapitre présente les modèles existants et décrit leurs principes et applications. Une méthode qualitative est développée puis comparée à d'autres démarches similaires ainsi qu'à des modèles empiriques afin d'étudier leurs performances dans la caractérisation ou la prédiction des expositions aux solvants de laboratoire (chapitre 2). Nous suggérons l'utilisation en première intention des méthodes qualitatives afin de définir les situations d'exposition prioritaires vis-à-vis desquelles les modèles empiriques peuvent secondairement servir à estimer un niveau d'exposition. Les modèles statistiques de régression linéaire sont étudiés à travers une application dans le secteur du compostage des déchets, confirmant leur capacité à caractériser des déterminants spécifiques de l'exposition (chapitre 3). Les modèles physiques mécanistiques sont testés afin de reproduire des profils temporels de concentration, puis utilisés de façon probabiliste pour estimer la distribution des expositions et des risques sanitaires (chapitre 4). Ces modèles permettent d'approcher la variabilité spatio-temporelle des expositions et d'identifier les mécanismes à l'origine des expositions. Les forces et limites de ces différentes approches sont comparées et des préconisations d'utilisation sont définies, assorties de perspectives de travail (chapitre 5). Mots clefs : exposition, risques, modèles, déterminants, prédictions, variabilité, incertitude. / The health risk assessment method involves the assessment of occupational exposure to pollutants. Restricted to inhalation exposure, several models can be used in order to either characterize or estimate exposure levels. The objectives are to study and compare the main existing models, and eventually to develop new ones, in order to help validating them and also to define their usefulness in addition to traditional individual monitoring. In a first chapter we present the existing models and describe their applications. A qualitative method is developed then compared to both other similar methods and empirical models in order to compare their performances in characterizing or predicting exposure to solvents in laboratories (chapter 2). From this work, we suggest the initial use of qualitative methods to define priority exposure scenarios, then the use of empirical models to predict inhalation exposure. Linear regression statistical models are studied in the field of waste composting, confirming their ability to identify specific determinants of exposure (chapter 3). Physical models are tested in order to reproduce observed time-varying exposure profiles, and then used to estimate the distribution of exposure and health risks (chapter 4). Such models are useful to describe the spatial and temporal variability of exposure, and help understanding the mechanisms of exposure. Strengths and weaknesses of all tested models are then compared and suggestions of use are made as well as work perspectives (chapter 5). Keywords: exposure, risks, models, determinants, predictions, variability, uncertainty.