Improving air pollutant exposure assessment by agent-based modelling – the example of nitrogen dioxide

Abstract

Agent-based models (ABMs) are a novel tool in environmental exposure assessment, simulating the mobility of people (represented by agents) within an environment and enabling individualised exposure modelling. This is particularly relevant for exposure to air pollutants with considerable spatiotemporal variation. In collaboration between the European Topic Centres on Human Health and the Environment (ETC HE) and on Data Integration and Digitalisation (ETC DI) of the European Environment Agency (EEA), we conducted an ABM feasibility study to assess the population's exposure to nitrogen dioxide (NO2) in Berlin, based on the MATSim framework. This simulation considers transportation patterns and individual daily activities, using an activity schedule scoring algorithm and detailed road networks. The study is based on a representative synthetic population sample that provides agents with socioeconomic information. By combining NO2 concentration maps and data, several scenarios were developed: a static "baseline" (agents at home all day), a "teleportation" scenario (agents switch instantly between home and work), and a "time-variant teleportation" scenario, incorporating typical NO2 level variation. The study reveals substantial exposure differences between the "baseline" and "teleportation" scenario of around ± 10% for approximately 20 % of the agents. Our study demonstrates the need for considering variation in NO2 levels for more realistic exposure assessments. ABMs offer more detailed ways of assessing air pollution exposure and health risks. By also considering socio-demographics, ABM results can provide more relevant information to policy-makers. Next steps include investigations into other pollutants, differences between population subgroups, and other urban areas in Europe. Quelle: Occupational Hygiene Society of Ireland (2026). Book of Abstracts: OHSI/ISES Europe joint 2026 Workshop; Dublin, Ireland, June 10th – 11th, 2026. Seite 76. https://ars.els-cdn.com/content/image/1-s2.0-S1438463926001008-mmc1.pdf (Zugriffdatum 01.10.2026)