Improving exposure assessment for nitrogen dioxide by integrating agent-based modelling

Background

Air pollution is a major environmental risk to population health. To date, most studies estimate the exposure based on air pollutant concentrations at the residential address, disregarding mobility. In this context, agent-based models (ABMs) can simulate the movement and interaction of individuals (agents) within an environment, allowing for a dynamic and individualised exposure estimation. This is particularly relevant for assessing the exposure to air pollutants with considerable spatiotemporal variability.

Methods

We conducted an ABM study based on a MATSim scenario to assess the population exposure to nitrogen dioxide (NO2) in Berlin. The study is based on a synthetic yet representative population sample that provides agents with socioeconomic characteristics and commuting patterns. The agent location data is combined with annual average NO2 concentration maps and hourly NO2 profiles from air quality monitoring stations. Several scenarios were developed: “baseline” (agents at home all day), “teleportation”, (agents move instantaneously between home and work), “time-variant teleportation” (incorporating typical daily NO2 level variation).

Results

NO₂ exposure between “baseline” and “teleportation” can vary up to ± 20% and is around ±10% for about 20% of the agents. For about 10% of all agents, daily exposure is estimated to be around 23-25 μg/m3, which is close to the WHO 24-hour guideline of 25 μg/m3 and well above the annual guideline value of 10 µg/m³. Peak traffic hours were found to be particularly relevant highlighting the importance of accounting for temporal variations in NO2 concentrations.

Conclusions

ABMs allow for more detailed assessments of air pollution exposure. By considering spatial mobility and socio-demographic factors, ABMs can offer policy relevant insights. This is a starting point for more realistic exposure and risk assessments for air pollutants and other determinants, with the potential to focus on vulnerable groups.

Key messages

• Agent-based modelling can support an improved assessment of air pollution exposure.

• Peak travel times and agents’ locations have a significant impact on exposure.