ENVESOME ABM

This record contains the agent-based modelling (ABM) software outputs reported in ENVESOME Deliverable D3.2, “Report on the Agent Based Modeling”, produced within ENVESOME — The Environmental Exposome and Health (Horizon Europe Grant Agreement No. 101157269).

The deposited outputs document the ENVESOME ABM for time-activity patterns, mobility and dynamic multi-stressor exposure across ten European Urban Living Labs: Aarhus, Athens, Barcelona, Berlin, Ljubljana, Łódź, Madrid, Porto, Thessaloniki and Vienna. The framework applies a harmonized computational workflow across all ULLs while using city-specific demographic, behavioural, spatial, transport and environmental input data.

D3.2 reports representative synthetic populations of 100,000 agents per ULL simulated over a continuous seven-day period at minute-level resolution. Agents are assigned demographic and socio-economic attributes, household and friendship networks, daily activity schedules, spatial activity locations and route-based mobility trajectories. Dynamic exposure is estimated for air pollution, noise and artificial light at night by intersecting agent trajectories with environmental stressor layers. A proof-of-concept food-consumption module is also included.

This archive preserves the graphical and tabular ABM outputs actually embedded in the supplied D3.2 deliverable. It includes 1,117 numbered figures, four numerical validation tables, and the image-based Athens noise-by-income output reported as Table 5. The outputs cover synthetic-population and activity validation, mobility, multi-pollutant and noise exposure, socio-economic gradients, social-capital analyses, hourly activity and exposure patterns, night-time light, food-consumption profiles, and environmental-justice analyses across the ULLs.

The quantitative validation outputs include activity-duration Kolmogorov–Smirnov statistics, χ² synthetic-population goodness-of-fit metrics, transport mode-share R² values and Wasserstein distances for spatial activity distributions. The combined validation results support comparative multi-city exposure assessment and structured subgroup comparisons.

Important limitations reported in D3.2 are retained in this deposit. Residential assignment is not explicitly conditioned on income or socio-economic status; indoor–outdoor attenuation is not currently represented; environmental monitoring density and pollutant availability vary between ULLs; and agents do not yet adapt their behaviour in response to environmental conditions. Absolute exposure estimates may therefore represent upper-bound personal-exposure estimates, while relative spatiotemporal, subgroup and cross-city comparisons are considered more robust within the current model configuration. VIIRS night-time radiance is used as a screening-level ambient proxy, and the food-consumption module does not yet calculate chemical contaminant intake.

D3.2 states that the ABM generates detailed minute-resolved per-agent output files containing activities, spatial coordinates and environmental exposures. Those raw agent-level files are not embedded in the supplied deliverable and are therefore not included or reconstructed in this archive. No raw agent data or plotted numerical datasets have been fabricated from the figures.