Future Ozone Exposure in Europe under Net-Zero Emission Scenario using Downscaled ICON-ART Simulations

Tropospheric ozone is a major air pollutant that poses significant risks to human health and ecosystems. As the European Union aims to achieve net-zero greenhouse gas (GHG) emissions by 2050, it is critical to understand how emission reductions will influence near-surface ozone concentrations. Unlike primary air pollutants, ozone is formed through complex, non-linear photochemistry involving nitrogen oxides, volatile organic compounds (VOCs), and meteorological conditions, making predictions of its response to emission reductions highly challenging.

In this study, we assess the impact of a transition to net-zero GHG emissions on near-surface ozone across Europe by comparing a reference year (2019) with a net-zero emission scenario for 2050. Our analysis is based on simulations with the atmospheric chemistry and transport model ICON-ART, which was specifically configured and further developed for air quality applications. The model incorporates the latest MOZART tropospheric chemistry scheme, enabling an accurate representation of key oxidation processes involving ozone, nitrogen oxides, and VOCs. ICON-ART further includes advanced modules for aerosol dynamics, gas-aerosol interactions, and emissions from biogenic and natural sources. Anthropogenic emissions are integrated via the ICON-ART online emission module. Together, the model components provide a physically consistent representation of regional scale atmospheric composition. However, the model's spatial resolution limits its direct applicability for exposure and health impact assessments.

To address this limitation, we apply a machine learning-based downscaling approach using the ensemble algorithm XGBoost to generate hourly near-surface ozone fields at 1 km spatial resolution. The model is trained on ground-based ozone observations and a comprehensive set of predictors, including ICON-ART chemical fields, meteorological variables, land use data, emission proxies, and topographic information. This hybrid framework combines process-based atmospheric modeling with a data-driven approach to capture fine-scale spatial and temporal variability in surface ozone. Moreover, the downscaling reduces model biases by leveraging observations to correct systematic errors in the ICON-ART outputs, improving accuracy and local representativeness.

Using the downscaled ozone projections, we examine changes in distributions, extreme events, and temporal dynamics between present-day and net-zero conditions. Our results provide new insights into how climate mitigation pathways may reshape ozone exposure across Europe and underscore the importance of high-resolution ozone projections for assessing the air quality implications of a transition to a net-zero society.